Background: We present an analysis of the impacts of heat and cold on all-cause mortality in 18 regions of metropolitan France between 2000 and 2010. Methods: Nonlinear distributed-lag models linking temperature percentiles and mortality were constructed for each region and then combined via a meta-analysis. These models were used to estimate the number of deaths attributable to heat and cold. A meta-regression incorporating environmental, demographic, and socioeconomic variables was also conducted. Results: Extreme heat (> 95th percentile) has a marked effect on mortality, with an immediate and rapid increase. Cold has an effect even at mild temperatures (< 25th percentile), with a gradual increase persisting up to 21 days after exposure. Across all 18 areas studied between 2000 and 2010, cold was responsible for 3.9% (95% CI: [3.2–4.6]) of mortality (impact over 0–21 days) and heat for 1.2% [1.1–1.2] (impact over 0–3 days). Conclusions: These results highlight the strong nonlinearity of the relationship between temperature and mortality, as well as the asymmetry in response between cold and heat, which must be incorporated into the development of prevention strategies.