A literature review was conducted to identify health effects attributable to low temperatures and those attributable to winter. Numerous studies highlight the existence of a V-shaped relationship between temperature and mortality, with a steeper slope on the high-temperature side than on the cold side. However, in absolute terms, cold weather is more deadly than heat. Mortality associated with cold thus extends far beyond the few annual deaths from hypothermia among the homeless. Seasonal variations with a winter peak are observed in several conditions, including coronary heart disease, strokes, and respiratory diseases: mortality increases almost linearly as temperatures drop. In January 1985, France experienced an unprecedented cold snap that led to significant excess mortality (+13%), primarily due to myocardial infarction (+17%), strokes (+54%), and pneumonia (+208%). Based on this review, the conditions of greatest concern in terms of public health appear to be cardiovascular diseases, respiratory infections and asthma, hypothermia, and carbon monoxide poisoning. For each of these conditions, particularly vulnerable populations have been identified and should be the subject of specific recommendations. However, questions remain regarding the links between cold weather and human health; these concern the choice of the best biometeorological indicator (absolute or perceived temperature, or other variables), the relationship between this indicator and the incidence of specific conditions, the population groups for which this relationship should be studied, the geographical specificity of these relationships, and finally the role of third-party factors: winter epidemics, air pollution, psychosocial factors, and the winter reduction in daylight and daylight hours. These questions warrant further studies, one of the objectives of which will be to determine whether cold-related illnesses are preventable and, if so, through which means of surveillance, early warning, and prevention. (R.A.)