Six major heat waves were recorded in metropolitan France from 1971 to 2003, each accompanied by a high excess mortality. The medical causes of death directly related to heat account for only part of this excess mortality, and almost all medical causes of death show an excess. Some populations – older people and subjects with certain specific pathologies – are particularly vulnerable. Mortality is also correlated with the “ordinary” daily fluctuations in summer temperatures. A model fitted for the period 1975-2003 is used to identify the temperature indicators that are strongly predictive of daily mortality. This model is then used to predict mortality during the summers of 2004-2006 on the basis of recorded temperatures in this period. It shows that the July 2006 heat wave was indeed accompanied by a large excess mortality, but that this excess was only one-third as large as could have been expected. The “vulnerability” of the French population thus seems to have been substantially lower in the July 2006 heat wave than it was in the period 1975-2003. Taken together these observations highlight the importance of heat-related mortality risks but also the potential capacity of populations to adapt to these risks, as was probably the case following the 2003 heat wave.