Objectives: From August 1st to 20th, 2003, the mean maximum temperature in France exceeded the seasonal norm by 11-12 degrees C on nine consecutive days. A major increase in mortality was then observed, which main epidemiological features are described herein. Methods: The number of deaths observed from August to November 2003 in France was compared to those expected on the basis of the mortality rates observed from 2000 to 2002 and the 2003 population estimates. Results: From August 1st to 20th, 2003, 15,000 excess deaths were observed. From 35 years age, the excess mortality was marked and increased with age. It was 15% higher in women than in men of comparable age as of age 45 years. Excess mortality at home and in retirement institutions was greater than that in hospitals. The mortality of widowed, single and divorced subjects was greater than that of married people. Deaths directly related to heat, heatstroke, hyperthermia and dehydration increased massively. Cardiovascular diseases, ill-defined morbid disorders, respiratory diseases and nervous system diseases also markedly contributed to the excess mortality. The geographic variations in mortality showed a clear age-dependent relationship with the number of very hot days. No harvesting effect was observed. Conclusions: Heat waves must be considered as a threat to European populations living in climates that are currently temperate. While the elderly and people living alone are particularly vulnerable to heat waves, no segment of the population may be considered protected from the risks associated with heat waves.