Apocalyptic visions of the environmental effects of nuclear war have been a part of our popular culture for decades. But apart from appreciating any entertainment value, the cognoscenti of nuclear war have regarded the doomsday predictions as ignorant at best, or dangerous propaganda at worst. The potential global environmental effects of nuclear explosions that were known before 1982—radioactive fallout and the destruction of the stratospheric ozone layer—were almost universally accepted in the strategic weapons community as being far short of true doomsday proportions. Indeed, for the combatant nations, such uncertain "secondary" effects were thought to pale before the assured direct effects of blast, heat and local radioactivity. From a scientific standpoint, this skepticism of environmental doomsday effects was probably justified in the sense that a large nuclear war would have been more devastating to the superpowers than any known indirect effects. The discovery of "nuclear winter" has challenged this skepticism because it has been much more compelling scientifically than the earlier predictions of global environmental effects. It has even been referred to as an inadvertent manifestation of Herman Kahn’s "doomsday machine."

The nuclear winter hypothesis, stated simply, contended that the smoke and dust placed in the atmosphere by a large nuclear war would prevent most sunlight from reaching the earth’s surface and produce a widespread cooling of land areas. The first two climatic conclusions of the theory were the most important: effects would be severe (weeks of sub-freezing temperatures), and effects would be widespread (at least hemispheric in scale). These grim scientific conclusions gave rise to two unique implications: the possibility of human extinction, and the potential suicide of an attacker even without retaliation by the attacked party. These implications, if confirmed, would indeed approach the definition of the traditional doomsday machine.

Another assertion was added to the hypothesis in the form of a scientific judgment: namely, that a "threshold" existed above which the climatic effects of a nuclear attack would become catastrophic. Thus, this doomsday machine did not possess a hair trigger, and would allow nuclear wars to be fought at some level substantially below the destructive potential of the current nuclear arsenals without global climatic catastrophe.

An additional major scientific conclusion closely followed the announcements of severe and widespread effects, but it initially received much less attention. Despite early suggestions that a nuclear winter was quite probable as long as a substantial number of large cities were attacked, many scientists concluded that the magnitude of effects would indeed be strongly dependent on uncertain, or even unknowable, factors.

The severe conclusions about nuclear winter provoked a broad spectrum of suggested responses for strategic policy. For many who took nuclear winter seriously, a perceived solution was a drastic reduction of nuclear arms to a level no greater than that necessary to constitute a minimal deterrent. On the other hand, the U.S. Department of Defense—which accepted the possibility of nuclear winter—argued that strengthening deterrence, combined with more research into nuclear winter, would be the best response. Thus, the potential for nuclear winter effects was used by the Department of Defense as another reason to support the continued modernization of U.S. strategic forces with smaller, more accurate warheads, and to pursue the Strategic Defense Initiative (SDI). Moreover, it was suggested that the ongoing modernization of strategic doctrine—expanded kinds of options for nuclear strikes and options for "limited" strikes—would reduce the probability of nuclear war by providing a more credible deterrent.

Cynics and agnostics were also well represented in the policy debate. Many considered the conclusions drawn from the nuclear winter theory, especially the early ones, to be too uncertain to provide the basis for any discussion of policy implications. Others pointed out that the horrible effects of nuclear war for combatant nations were already widely accepted, that any environmental or indirect effects on noncombatants would not further motivate the superpowers to seek arms control, and that additional bad effects would only add marginally to an already strong mutual deterrence.

We intend to show that on scientific grounds the global apocalyptic conclusions of the initial nuclear winter hypothesis can now be relegated to a vanishingly low level of probability. Thus the argument that nuclear winter provides the sole basis for drastic strategic arms reductions has been greatly weakened. But, at the same time, there is little that is thoroughly understood about the environmental effects of a nuclear war. In particular, we do not think that all environmental effects should once again be considered as "secondary." Important environmental and widespread societal effects of nuclear war remain quite probable and do suggest further scientific and policy considerations. Our current understanding of environmental effects will be reviewed and then used to bolster arguments for strengthened strategic stability, not necessarily excluding newer strategic systems, but at significantly reduced levels of arsenals.


It is reasonable to ask why the scientific basis of the theory of nuclear winter still provokes such divergent scientific opinions. The answer involves both the complexity of the problems and the severity of predicted effects. It is important to remember that the widespread radioactive fallout and ozone effects, although substantial, were never really thought by knowledgeable researchers to have a doomsday potential. In contrast, the original nuclear winter results showed truly catastrophic consequences arising from general war scenarios in the best available calculations. Thus, this new hypothesis could not be readily dismissed. Moreover, the problem of nuclear winter involves more scientific disciplines and more crucial areas of uncertainty than the earlier environmental problems. In particular, estimates of smoke production cannot be made from old nuclear test data and cannot be well bounded theoretically.

Initially, the scientific basis of nuclear winter rested exclusively with the TTAPS group and their first calculations. After the original discovery that smoke could pose the most serious environmental threat of a nuclear war, the TTAPS group began to use a computer model to study the effects of war-generated smoke and dust on the earth’s climate. The model was one-dimensional; that is, it did not take into account north-south and east-west directions, but instead treated the earth as a homogeneous all-land sphere having a temperature that depended only on the up-down direction (atmospheric altitude). Thus, the model had no geography, no winds, no seasons, instantaneous spread of smoke to the hemispheric scale, and no feedback of atmospheric circulation changes on the rate of smoke washout by rainfall. Despite these limitations, the TTAPS calculations did offer state-of-the-art estimates of the sunlight and infrared radiation absorption of a given amount of smoke and dust in a vertical column of the atmosphere.

The TTAPS study included numerous war scenarios that were used in simulations of climatic effects. The curve marked in Figure 1 (p. 994) shows the time evolution of land surface temperature produced by the TTAPS nominal, or "baseline," case in which a 5,000-megaton nuclear war was estimated to produce 225 million metric tons of smoke and 65 million tons of small-size stratospheric dust. Many other war scenarios generated catastrophic coolings of varying severity, but the TTAPS authors responsibly cautioned that the weaknesses of the model would have to be addressed before the calculations could be treated as conclusive for any given war scenario.

By the summer of 1983 the major areas of uncertainty in the nuclear winter problem were generally recognized even though public announcement of the theory was months away. Since that time increasing numbers of researchers have been steadily improving our understanding of these critical areas. Excluding various war scenarios, the most crucial nuclear winter uncertainties were thought to be: (1) the amount and "blackness" of generated smoke, its initial altitude distribution and its removal by rainfall in clouds associated with the fires; (2) the degree to which smoke might spread globally and how fast it would be further removed from the atmosphere; and (3) the detailed regional and seasonally dependent pattern of weather and climatic effects. Significant progress has now been made in research in each of these areas, especially the prediction of global climatic effects given a scenario of smoke in the atmosphere.


As noted above, one of the most controversial concepts to emerge from nuclear winter research was the notion that a threshold amount of smoke could trigger catastrophic climatic effects. This notion was based on a number of simulations in the TTAPS study, but was primarily founded on a hypothetical "100-megaton war" in which 100 major cities were targeted exclusively. The smoke so generated was sufficient to cause northern hemispheric land surface temperatures to drop well below the freezing point in the one-dimensional model. Ignoring for the moment the question of whether such a "city-busting" war scenario is plausible, we can use it to illustrate why the strategic implications of the threshold concept caused it to be widely debated.

One bizarre implication of the threshold theory was that a sharp threshold might tempt an attacker to use up the entire "quota" of smoke production, leaving little alternative to the side considering retaliation but to trigger nuclear winter or do nothing. Furthermore, it was suggested that a well-defined threshold might encourage the fighting of "small" nuclear wars.

In response to these sorts of criticisms Carl Sagan replied, quite sensibly, that it would be dangerous to gamble that the spiral of escalation in the wake of a sub-threshold attack would remain safely below the level necessary to create serious climatic effects. Indeed, this danger is what led to the most visible and controversial aspect of Sagan’s strategic policy conclusions: the imperative for reducing the world’s nuclear arsenals to a level safely below that sufficient to trigger nuclear winter. Given that the total yield of 100 megatons could trigger nuclear winter, the implication was that the level of superpower arsenals would have to be reduced to about one percent of their current value. (We will refer to this degree of nuclear arms reductions as "drastic cuts.") Thus, the call for a policy of drastic cuts in strategic arsenals was directly linked to the threshold concept.

If one accepts the idea of a threshold for nuclear winter for the sake of strategic argument, then the question of how to quantify it becomes important. Given the substantial uncertainties in defining a threshold, and the potentially grave consequences of exceeding it, the argument was made for deriving a threshold based on a worst-case smoke-producing assumption of targeting. Taken in this light, the strategically dubious "100-megaton war" can be seen as a worst-case assumption serving to delineate a prudent threshold. The 100-megaton scenario used 1,000 warheads, but it was never suggested that the threshold would be very well defined. For example, in his Foreign Affairs article, Sagan set a crude threshold of around 500 to 2,000 warheads, primarily attributing the uncertainty to assumptions of targeting and weapon yields.

Ironically, just when the strategic implications of the threshold concept were starting to be debated in the strategic policy community, the strongest scientific arguments against the concept emerged. To understand why the threshold concept is not scientifically persuasive, one must be aware of the limitations inherent in the models that do not take account of geography.

It is tempting to interpret a calculation in which surface temperature is represented as a single global number in terms of a dramatic physical threshold, e.g., the freezing point. But such calculations cannot capture the true geographical and seasonal heterogeneity of climate. A global model developed at the National Center for Atmospheric Research studied the inclusion of north-south and east-west dimensions and how such improvements would modify the results of one-dimensional models.

We found that the oceans, with their vast storage of heat, would reduce the magnitude of average continental cooling by a factor of two in the summer, compared to the cooling calculated by assuming a land-covered planet. The estimated cooling effect in winter was smaller by a factor of ten than the TTAPS annual estimate, because northern hemisphere mid-latitude land areas are already cold in winter. Even when we assumed a uniform smoke cloud to exist over the middle part of the northern hemisphere, the surface temperature reduction was unevenly distributed—much less along western coasts and even more than one-dimensional model results in some mid-continental cold-weather fluctuations in summertime.

In short, simulations using geographically realistic models produced such a wide range of consequences for any given war scenario that it became clear that the elegant and strategically compelling idea of a threshold was an artifact of a simplified model. This observation remains true. Hence, it is questionable to predicate any strategic policy options on the existence of a nuclear winter threshold, even a "fuzzy" one.


Although the bulk of the news media coverage of the nuclear winter debate in late 1983 concentrated on the more dramatic conclusions and criticisms of the theory, there was some press attention to the increasingly complex scientific research efforts that were under way. As these scientific efforts intensified in 1984 and 1985, however, popular interest in nuclear winter appeared to have decreased. On the other hand, interest in the scientific community picked up momentum as the issues became more complex and thus more scientifically exciting.

In 1982, after the discovery of the smoke problem, an international group of scientists began to plan a major study of the environmental consequences of nuclear war under the auspices of the Scientific Committee for Problems of the Environment (SCOPE)—a subgroup of the well-respected International Council of Scientific Unions. The SCOPE findings were released in two volumes in September 1985, the first covering physical effects and the second biological and other environmental effects. Also in 1982, a group of U.S. scientists at the National Academy of Sciences (NAS) realized that a reassessment of the atmospheric effects of nuclear war was needed, particularly since a previous academy study published in 1975 ignored the importance of fires and smoke. This new NAS study, co-chaired by Harvard’s George Carrier and retired Vice Admiral William Moran, was released in December 1984.

Both studies of physical effects stressed the same two themes. First, that there remained great uncertainties—some that could never be resolved—over every link in the chain of phenomena leading to the two roughly defined phases of environmental effects: "acute" (one to 30 days) and "chronic" (months to years). Second, despite the cascading uncertainties, both reports concluded in strong language that very large climatic effects were possible and should not be ignored. Not surprisingly, these scientific committees emphasized that accelerated scientific research to reduce the uncertainties should be a high priority.

It is noteworthy that both studies examined war scenarios that were not based on either worst-case assumptions or most likely scenarios. Instead, "baseline" cases were considered that were believed at the time to be plausible examples of a large general nuclear war. The NAS committee in 1983 adopted a baseline scenario that placed 180 million metric tons of moderately dark smoke in the atmosphere within a few days of the start of a 6,500-megaton nuclear war. This scenario became the basis for many subsequent analyses of nuclear winter through 1985. In a sense, these reports—especially NAS which was available months earlier than SCOPE—helped to legitimize nuclear winter as a scientific research topic in general, and the use of mathematical climate models as appropriate tools for prediction in particular.

The SCOPE effort went further than the NAS report because it included a study of biological effects. The U.S. interagency research effort in nuclear winter deliberately downplayed biological research as premature, given the many uncertainties in the physical sciences still to be resolved. Nevertheless, consideration of the biological consequences was appropriate for two basic reasons. First, if a global "deep freeze" scenario is considered, then it is rather obvious that the biological consequences would be catastrophic to both natural and agricultural systems worldwide. But by late 1984 it was becoming increasingly apparent that such global freeze scenarios were of exceedingly small probability, and that the response of vegetation to coolings of less intensity and shorter duration needed to be assessed. This, of course, is a more challenging scientific problem than simply observing that everything would freeze; thus, assembling an international group of biologists to think about such issues was sensible.

The second reason to consider biological consequences before all the physical facts were confirmed is simply that it is important for physical scientists to know what variables are of greatest importance to estimating biological damages from low temperatures, radioactivity or other factors. Therefore, having some idea of what was important to the biologists could help the physical scientists choose their approaches to atmospheric research questions.

Finally, the SCOPE biological report was important for what it did—and did not—say. It did not discuss the plausibility of human extinction as a result of nuclear war, thus implicitly rejecting the notion that extinction was a "real possibility." It did, on the other hand, follow up on the indirect global societal effects of nuclear war that had been known for decades but never treated in a very quantitative manner. For example, the SCOPE biologists calculated that at least hundreds of millions of people could die of starvation in noncombatant nations from disruption of food trade alone, even if no smoke, dust or radioactivity entered their territories. Any comprehensive assessment of the indirect consequences of nuclear war clearly needs to consider not only environmental effects, such as atmospheric changes and radioactive fallout, but disruptions of basic societal functions as well (e.g., trade in basic commodities such as food, fertilizer, medicine, spare parts and fuel). Thus, even if further research substantially reduces the probability of environmental effects, the examination of global societal effects undertaken by the SCOPE biologists should be a very high priority for further study.

Let us sum up: despite the continued potential for serious nuclear winter effects, there does not seem to be a real potential for human extinction; nor is there a plausible threshold for severe environmental effects. Thus, the two unique conclusions of the original nuclear winter idea with the most important implications for policy have been removed.


There remains the question of the extent to which the indirect environmental effects will be significant compared to the direct effects of nuclear war. To make such an assessment requires an accounting of the state-of-the-art research on nuclear winter. The following discussion describes our current understanding of the physical effects of a nuclear winter, a substantial portion of which has been developed after the publication of the most recent general assessments, the SCOPE reports, which include material up to the summer of 1985.

Given a plausible war scenario, perhaps the single most uncertain area of the nuclear winter problem is the determination of how much smoke would be produced, how "black" the smoke would be, and how much would be rained out immediately. Studies are in progress now to narrow these uncertainties. One recent estimate indicates that the amount of smoke initially produced in a major nuclear war may have been overestimated by roughly a factor of two to four in the NAS report. The blackness of the smoke (determined by the percentage of black carbon soot in the smoke), however, can be just as important as the amount of smoke in determining atmospheric effects. Indeed, an effective amount of smoke can be defined as the product of the total amount of smoke and the smoke blackness. It is possible that current appraisals underestimate the blackness of typical smoke from urban fires. This potential error could substantially compensate for a downward revision in estimates of total smoke produced.

Potential rainout of smoke near the fires (like the "black rain" that followed the atomic bombing of Hiroshima and the subsequent conflagration) controls the amount of smoke that can eventually spread and cause climatic effects. The estimates of rapid rainout of smoke in large thunderstorms associated with city fires are very problematic, but ambitious recent calculations indicate that even under ideal rainout conditions a substantial amount of smoke would probably remain in the atmosphere.

State-of-the-art assessments of the climatic effects of nuclear winter are based on results from fully three-dimensional models that simulate the evolving patterns of global weather on an earth having realistic geography. The simulations recently performed at the National Center for Atmospheric Research have used a model which resolves the atmosphere from the surface to about 30 kilometers altitude (the middle stratosphere) and which resolves geographic features at about 5° latitude by 7° longitude resolution. The model includes transport by winds, removal of particles from the atmosphere by rainfall and other processes, and detailed calculations of sunlight transmission and infrared "greenhouse" effects—all for both smoke and dust, although smoke is by far the larger contributor to climatic effects. In contrast, recall that the TTAPS calculations had no geographic resolution, and hence could not adequately predict the global spread or removal of smoke from the atmosphere. Of course, even the most sophisticated models include simplifying internal assumptions; thus we anticipate that present quantitative results will change somewhat as improvements continue to be made.

In the recent NCAR research, we have not adopted any particular detailed war scenario other than the obvious assumption that the smoke and dust would come from the NATO and Warsaw Pact countries. In fact, there is no general consensus on the amount and blackness of smoke that would exist in the atmosphere a few days after the start of a major nuclear exchange like the 6,500-megaton war the NAS study used as a baseline. Given this large range of uncertainty, we have used three different amounts of moderately black smoke—20, 60 and 180 million tons—to bracket what is currently thought to be a reasonable range of smoke amount and blackness for a large nuclear war. It should be pointed out that the NAS baseline smoke amount estimate (180 million tons) now appears to lie closer to the plausible upper limit of effective smoke amount than it once was thought to. The war is assumed to take place during a typical day in July, with smoke generated for the first two days.

The land surface temperatures produced by the three smoke cases are shown in Figure 1. This figure shows that the average temperature changes for the northern hemisphere mid-latitudes are considerably smaller than the original estimates of one-dimensional models, and are about two-thirds of the temperature changes found in our original three-dimensional calculations. These temperature changes more closely describe a nuclear "fall" than a nuclear winter.

The reasons for the moderation of temperature compared to the original calculations are well understood: first, the oceans have a large heat capacity, which ameliorates the cooling over land. Second, about three-fourths of the smoke is removed from the model’s atmosphere over the course of 30 days. Third, the infrared "greenhouse" effect of the smoke, which was not included in earlier three-dimensional models, does produce a significant mitigation of the surface cooling. We must stress that our results are for July, the month in which the temperature changes are likely to be largest. Similar calculations for January show much less effect simply because at that time it is already winter in the northern hemisphere.

??NOTE: The three curves denoted as "NCAR" give the average July surface temperature of all land areas within the 30-50°N latitude zone for the 30 days following a hypothetical one-day nuclear war in which three different amounts of moderately black smoke are injected into the atmosphere over the course of the first two days. "Tg" refers to teragrams, or million metric tons, of smoke. (A metric ton is 2,200 pounds.) Temperatures at individual geographic locations and times can be more or less severe than the averages shown. The normal July surface temperature and the freezing point of pure water are shown as the thin horizontal lines across the graph. "Wiggles" in the curves are a consequence of averaging over weather variability, which the model simulates. The result of the first 30 days of the TTAPS model baseline case (see p.985) is shown as the dotted curve and is given for reference only since it is difficult to compare directly the NCAR results with the TTAPS results. The TTAPS model predicted an average northern hemisphere temperature for an all-land planet for annually averaged conditions. One difference between annual average and July cases can be seen in the colder Day 0 starting point for the TTAPS curve.

The curves in the figures represent averages over all the land areas in wide latitude zones. At any specific location in the model, however, the temperatures are considerably more variable. For example, some large areas in the interiors of the North American and Eurasian continents, particularly in Canada and Siberia, fall below freezing intermittently in the two cases of larger amounts of smoke. On the other hand, some areas near coasts experience little effect. Thus, it can be misleading to interpret the curves on the figures without taking into account geographic and weather variability as well. Indeed, for certain biological impacts it would be sufficient to have only a few hours of temperature below some critical level—e.g., subfreezing for wheat, or 10-15°C (50-59°F) for rice.

??NOTE: This figure is similar to Figure 1, except that the curves give land surface temperatures for three adjacent latitude zones in the northern hemisphere for the case in which 180 million metric tons (teragrams, Tg) of smoke are produced in July. The black dots at the start of each curve indicate the normal temperature of each latitude zone before the addition of smoke to the atmosphere. Compare the smaller drop (about 5°C) in the curve for latitude 10-30°N and the larger drops (10-15°C) in the other two curves, indicating the smaller cooling effect for the subtropical areas.

The global spread of acute climatic effects is assessed in Figure 2. Previous atmospheric circulation models have supported the notion that smoke would spread upwards and across the equator into the southern hemisphere for a war taking place in the northern hemisphere during the summer. While this has not changed, our current estimate is that the amount of smoke thus transported would be relatively small. This is reflected in the temperature curve for the subtropical latitude band 10-30°N, which shows only a small average effect, even for our largest case of 180 million tons of smoke. Equatorial and southern hemispheric temperature effects are very small in all these cases. The much greater severity of effects at high northern latitudes, however, is shown by the results for 50-70°N.

Since smoke is continuously removed from the atmosphere in these simulations, an interesting question to ask is whether one could minimize nuclear winter effects by fighting a war slowly enough that a damaging amount of smoke could not build up in the atmosphere. This idea is not merely academic since the notion of being prepared to fight a "protracted" nuclear war has become a topic of some debate in the past decade. We have examined this idea by assuming a ten-day war that generates the same total amount of smoke as the one-day war. The results of the case using 60 million tons of smoke are shown in Figure 3. The steady input of smoke in the protracted war case actually produces more cooling after a week than the rapid war case. The implication from this simulation is that fighting a large war slowly (in the unlikely event that such a thing is possible) would not necessarily ameliorate environmental effects.

??NOTE: The land surface temperature of the 30-50°N latitude zone for two cases having different assumed war durations, each of which produces a total of 60 million tons (teragrams, Tg) of smoke. The "one-day" war, allowing for some delay in fires, actually produces smoke over two days. The "ten-day" protracted war produces smoke continually over ten days, but at a lower rate than the one-day case. Despite the fact that smoke is continually being removed in each case, the slower rate of smoke input in the ten-day case actually worsens the climatic effects when compared to the faster war.

The difficult issue of chronic climatic effects (months to years) has not received nearly as much attention as the acute nuclear winter effects (one to 30 days), but will undoubtedly assume greater research importance in the future. A substantial fraction (perhaps ten to 30 percent) of smoke injection from any massive nuclear war would likely reside in the stratosphere for months before finally being removed from the atmosphere. Although the obscuration of sunlight caused by such a shroud would not be sufficient to cause severe surface cooling, it could create other serious climatic effects, e.g., anomalous late spring and early autumn frosts, or a disruption of normal monsoons and summer rainfall over continental areas. Such effects could greatly hamper agricultural recovery in those areas still having sufficient social stability and economic resources to carry out viable agricultural practices.

The previously discovered chronic environmental problems of global radioactive fallout and ozone layer depletion continue to be studied. It now appears that the trend to smaller warheads has made the intermediate-term (days to weeks) and global fallout problems somewhat more serious, but the ozone problem has become less serious. As noted earlier, neither problem considered separately is thought to pose a global threat that is substantial compared to the direct effects of a nuclear war. But considering all the chronic, indirect effects of a large nuclear war separately may be a misleading exercise; all the chronic effects would, to some degree, act synergistically with each other, and with the direct weapons effects, to produce unprecedented worldwide human misery.

The interaction between a failure of the Asian summer monsoon and the disruption of international trade in food, for example, would be a potent prescription for mass starvation on the Indian subcontinent. Many other interactions can be speculated upon, such as those of radiation-induced depression of the human immune system, disruption of medical services and epidemics of contagious diseases; such synergisms deserve the serious study that might allow us to draw less speculative conclusions. Therefore it is still quite plausible that climatic disturbances, radioactive fallout, ozone depletions and the interruption of basic societal services, when taken together, could threaten more people globally than would the direct effects of explosions in a large nuclear war.

In assessments of the severity of the nuclear winter problem, the long-term trend in temperature estimates has been directed away from the most severe effects. This trend has not been a smooth one, however. Researchers have made refinements and enhancements to the theory that have made the climatic effects both more and less severe. This was expected because original estimates of uncertain effects tried to incorporate "middle of the road" assumptions. Thus, researchers knew that the tentative conclusions about nuclear winter would change with time depending on the latest refinements. This is, of course, a legitimate and normal process of scientific assessment. Despite the potential for future changes in conclusions, we think it is unlikely that some unforeseen effect would either bring the estimates of nuclear winter back to the range of total climatic catastrophe or eliminate environmental effects altogether.

The cases we have examined (20, 60 and 180 million tons of smoke) bracket what we believe to be a wide range of probable effective smoke amounts for a large nuclear war. But it is conceivable that even larger amounts of smoke could be created, although the probability now seems small. Similarly, less than 20 million tons of smoke could be produced, at some low probability, even after a large war in which many cities were struck. In the absence of near-infinite consequences (e.g., human extinction) we believe it is unwise to base expensive policy decisions (e.g., SDI) solely on either of these very low probability cases.

Despite the downward trend in assessments of the effects of nuclear winter, there are good reasons to continue the research effort. First, it is essential that, to the extent possible, decision-making be placed on an increasingly firm factual basis. Second, as noted, assessments of smoke amounts do change and must be tracked, for example, by an ongoing parallel research program to improve atmospheric models. Third, the nuclear winter problem has strong links to several other atmospheric environmental problems (e.g., acid rain and the carbon dioxide "greenhouse effect"), and has contributed to improving our understanding of the behavior of the atmosphere. Finally, and perhaps most importantly, the significant environmental and indirect effects that remain highly plausible must be more carefully considered.


Our current understanding of the climatic effects of nuclear war has changed the way that many previously proposed policy implications of nuclear winter should be viewed. Indeed, several of the strategic implications of nuclear winter, as it was originally conceived, are no longer justified: for example, the anxieties about a "threshold" for severe effects and the notion that climatic effects alone would serve as an in-kind retaliation against a major preemptive strike. The idea of automatic suicide is now unsupportable given that a scenario of weeks of continuous subfreezing temperatures on a continental scale is no longer plausible. However, the substantial amounts of smoke, dust and radioactivity that would necessarily be injected into the atmosphere in any large-scale preemptive strike should still add some measure of deterrent value, even if short of assured national suicide.

Apart from biologically oriented environmental effects, atmospheric smoke and dust could affect strategic surveillance, early warning, missile defense, communications, and attack assessment during a nuclear war. The primary effect would be obscuration of visible and infrared light that could, for example, degrade the information available from satellites or hinder the atmospheric propagation of laser beams used for communications or defense. It is arguable whether potential obscuration effects are strategically important because such effects would particularly apply to operations that would be carried out very early in a general nuclear war. Although there is debate on the issue, and the U.S. strategic command, control and communications (C3) system is currently undergoing significant upgrading, the present informed judgment is that strategic C3 could be maintained for only a short time if a massive attack were directed at the C3 facilities themselves —probably a shorter time than it would take obscuration to become a serious problem.

In summary, obscuration effects would not likely be as important in a small protracted war (if such a thing is possible) as in a large general war, and they may simply be irrelevant in a conflict in which C3 and missile defense systems were attacked early. But since the effect of obscuration depends on the uncertain details of how a nuclear war would be fought, obscuration itself adds yet another uncertainty that may add some measure of additional deterrent.

There are policy changes that could minimize all potential effects related to nuclear winter, but we believe that these should also be judged by other criteria. For example, drastic cuts (by 90 to 99 percent) in the world’s strategic arsenals would likely reduce the threat of environmental effects, but such a policy also has serious drawbacks. For example, verification of small numbers of weapons would be difficult, and "break-out" growth to larger arsenals would be a perceived threat. A small deterrent force might also be misread as a "weak" deterrent, thus reducing the inhibitions against a preemptive strike in times of crisis. In addition, horizontal nuclear arms proliferation could be encouraged, as some countries could achieve nuclear superpower status with a much smaller effort; one result of this could be even more difficulty in reaching arms control agreements. Lastly, it is possible that given a relatively limited number of weapons, one would choose to announce publicly a primarily countervalue (and thus smoke-producing) targeting policy to maximize the deterrent threat of one’s strategic force. Nuclear winter, clearly, is not valid as a sole basis for embracing drastic cuts in nuclear arsenals.

Of course there are technical "fixes" as well that could conceivably minimize both unintended direct and indirect effects of nuclear weapons. For example, the deployment of low-yield earth-penetrating warheads would greatly reduce the potential for large-area fires and smoke generation. Such a change in arsenals would have to be defended against the concern that the modernization effort would make a nuclear war less catastrophic, therefore more "thinkable." More important, arguments for the development of any new generations of strategic weapons, which sometimes appear valid in isolation, can in effect be arguments against applying brakes on the strategic arms race and nuclear proliferation. These brakes include both arms limitation and testing restraints, such as the SALT II ban on more than one "new" ICBM, and the frequently discussed Comprehensive Test Ban Treaty.

Wide-area non-nuclear strategic defense, if it were possible, could reduce many effects of a nuclear war, but it is the subject of intense controversy in its own right. It seems likely that the antimissile systems that would be deployed first would be to protect point targets (e.g., missile silos) since that defense task is much easier than defending something as large and "soft" as a city. In this case, as in the case of drastic cuts in the arsenals, an adversary may choose to shift targeting toward cities in an attempt to maintain a credible deterrent, which in turn increases the risks of nuclear winter. Alternatively, an adversary might seek to overwhelm the defense by increasing offensive deployments, which also raises those risks. But in the absence of such changes in offensive strategy—and given our current understanding of environmental effects—concerns about nuclear winter add only a small component to the overall SDI debate.

There are actions that could be taken to reduce the environmental effects of nuclear war that, in principle, need not be very controversial. For example, both sides could minimize potential indiscriminate environmental effects by placing targets with a high fuel density into "withhold" categories in their strategic war plans. A counter to this commonsense proposal would have the superpowers place above-ground oil reserves near strategic targets, thus insuring a nuclear winter if one side attempted a counterforce attack. But the supposed assured retaliation afforded by such an improbable "doomsday" machine, even if pursued on a massive scale, might simply make the effects in the attacked nation worse without creating suicidal effects for the attacker.

Reducing the total explosive megatonnage of the world’s strategic arsenals would also lessen the chances of global environmental effects should a nuclear war be fought. It is true that the trend of the past two decades has shown a reduction in the total yield of the U.S. arsenal, but future plans, such as the deployment of the next generation of D-5 missiles on Trident submarines (nearly 2,000 megatons of explosive equivalent on about 20 submarines), and to a lesser extent the deployment of MX missiles, do not bode well for the continuation of this trend. In any case, a more fundamental consideration of strategic arms deployments should be the need to stabilize and strengthen deterrence by avoiding systems that are tempting targets for a preemptive strike by virtue of their poor survivability and obvious utility for a disarming counterforce attack.


Official U.S. responses to the nuclear winter theory went through early phases of disarray and cynicism, to settle eventually on qualified acceptance and a low-level interagency research program overseen by the Office of Science and Technology Policy, with highly focused research efforts sponsored by the Defense Nuclear Agency, Department of Energy and the National Science Foundation. Relative to official U.S. reticence, the Soviet Union has, to many Western observers, appeared steadfast in its ostensible acceptance of the absolute horrors implied by the earliest nuclear winter scenario. Remarks by Soviet spokesmen have especially emphasized the idea that nuclear winter "retaliation" would be automatic following a nuclear strike by either side. As one of the milder examples, Sergei Kapitsa, chief of department at the Institute of Physical Problems of the Soviet Academy of Sciences, remarked in the October 1985 issue of the Bulletin of the Atomic Scientists that "even if one power strikes, the consequences will be global and retaliation automatic."

The Soviet position, however, is frequently believed to have been adopted for its propaganda value, presumably in the hope that U.S. and West European public opinion could be used to pressure the United States into arms control negotiating positions more favorable to the Soviets. Soviet research on nuclear winter, at least that which has been made available to the West, has not up to now been as helpful and forthcoming as we would have hoped. Some limited fire data have been received, and some Soviet scientists did participate in the international SCOPE assessment. For climate estimates, Soviet scientists have typically used war scenarios and other parameters taken from U.S. research and applied them to rather less capable computer models.

It is well known that differing perceptions of technical and political problems often contribute to the difficulties the superpowers have in coming to agreements on strategic issues. If environmental effects were perceived to be extreme enough for some sort of policy action other than research, then a response to nuclear winter could face the problem of asymmetric superpower perceptions. For example, if neither side believed in nuclear winter, despite any scientific evidence to the contrary, then the problem of mismatched responses would be moot. Similarly, if both sides believed comparably, then they would have some common ground for working out mutually satisfactory responses to the perceived threat. Problems arise, however, if one side perceives nuclear winter to be a genuine threat, but the other side does not.

One of the two possible cases of asymmetric perception seems implausible, namely the situation where the United States does not consider nuclear winter to be significant and the Soviet Union does. This is unlikely for at least two reasons. First, most of the progress in evaluating climatic effects of nuclear war has come from unclassified work by U.S. scientists, both in and out of government, and will probably continue to. It seems doubtful that the Soviets would sincerely embrace U.S. research if the U.S. government did not. Second, the Soviet Union has never publicly displayed much appreciation for "finesse" in strategic doctrine, for example, war-fighting strategies, limited options and the like. Therefore, a unilateral Soviet response to a less-than-apocalyptic nuclear winter threat would be out of character since such a response would probably entail modifications of the Soviet arsenal and doctrine.

The last situation—the United States believes in nuclear winter and the Soviet Union does not—would be particularly dangerous. It now appears that the U.S.S.R. has considerably more potential than the United States to unilaterally cause environmental damage from smoke due to nuclear war. The primary reason for this is that one estimate shows there to be more than twice as much burnable material in urban areas of the NATO countries than in the Warsaw Pact countries, according to the work of George Bing cited above. A secondary reason is that, since smoke production may be roughly proportionate to megatonnage (everything else being equal), the U.S.S.R. currently has about twice as much strategic megatonnage deliverable on short notice.

Furthermore, being in a position of publicly believing the theory could force the United States to try to adjust to the perceived threat of nuclear winter by changing arsenals and doctrine (while the Soviet Union did not), or else face public pressure for not acting despite a professed belief in the theory. The best hedge against the problems that can arise from differing perceptions of nuclear winter is to keep the results of all the nuclear winter research programs of both sides as widely available as possible.

Apart from such hypothetical asymmetric perceptions of the nuclear winter problem, there are real, physical asymmetries between the superpowers regarding potential climatic effects. It is rather ironic, given the Soviet Union’s greater capacity to produce smoke by attacking NATO countries, that advanced simulations show that the more poleward position of the U.S.S.R. and the greater landmass of the Eurasian continent put the U.S.S.R. at a disadvantage vis-à-vis the United States in terms of climatic effects. This is true for the same reasons that the U.S.S.R., on the whole, experiences more severe normal winters than does the United States.

Canada, having a large landmass at high latitudes, would face nuclear winter effects similar to those of the Soviet Union. This potential has not been ignored by the Canadians, and they have responded by performing their own assessment of the problem. Apart from this, and despite the substantial implications that environmental and other indirect effects hold for noncombatant nations, there has been surprisingly little international political interest in the subject. For example, the review of the Nuclear Non-Proliferation Treaty in the autumn of 1985 went smoothly despite the fact that the nuclear winter theory could have been used to accuse the superpowers of acting in bad faith by threatening noncombatant nations.

We cannot pretend to know what the legacy of the nuclear winter hypothesis will be. Perhaps it will be a scientific reconfirmation that the effects of a large nuclear war would not merely be scaled-up versions of effects from a war fought with conventional weapons. In the final analysis, though, we must recognize that whatever our level of understanding of the effects of nuclear weapons, and whatever our ability to apply technical fixes to weapons, defenses or doctrines, the problem of avoiding nuclear war is not amenable to scientific solution. This problem arises more from political differences than from the latest technical capabilities. If nuclear winter has made us more aware of the urgent need to find political solutions to the arms race and the threat of nuclear war, that alone will have made the entire exercise worthwhile regardless of the scientific disposition of the remaining uncertainties.

You are reading a free article.

Subscribe to Foreign Affairs to get unlimited access.

  • Paywall-free reading of new articles and a century of archives
  • Unlock access to iOS/Android apps to save editions for offline reading
  • Six issues a year in print, online, and audio editions
Subscribe Now
  • Starley L. Thompson is an atmospheric scientist and climate theorist at the National Center for Atmospheric Research (NCAR) in Boulder, Colorado. Stephen H. Schneider, an atmospheric scientist and public policy analyst, is Deputy Director of the Advanced Study Program at NCAR. Any opinions, findings, conclusions or recommendations expressed in this article are those of the authors and do not necessarily reflect the views of the National Science Foundation, which sponsors NCAR, or the Defense Nuclear Agency, which supported a portion of the research described here. The authors acknowledge the scientific contributions of Curt Covey, Filippo Giorgi and V. Ramaswamy. Copyright © 1986 by Starley L. Thompson and Stephen H. Schneider.
  • More By Starley L. Thompson
  • More By Stephen H. Schneider