Santé publique France has established a network to monitor the relationship between temperature and mortality based on multicenter time-series studies in 18 French urban areas. Its initial research focuses on the impacts of heat and cold over the 2000–2010 period. The relationship between temperature and mortality is highly consistent across the areas studied. Cold has a gradual effect, spread over several days, and is visible at mild temperatures (below the 49th percentile). Heat has a very rapid effect, concentrated in the first few days, possibly followed by a harvesting effect. Above the 99.2nd percentile, the immediate increase in mortality is only partially offset by the harvesting effect, which disappears entirely at percentiles above 99.8. For an extreme temperature (99.9th percentile), the cumulative mortality over 21 days is 1.96 times higher (relative risk (RR): 1.959 [95% CI 1.587–2.418]) than for a median temperature. Across all 18 urban areas, between 2000 and 2010, cold weather accounts for 3.9% [3.2:4.6] of mortality, and heat for 1.2% [1.1:1.2] when considering the overall impact, and 0.5% [0.4:0.6] when accounting for the harvest effect. The impact of cold stems from low relative risks (RRs) occurring on a significant number of days per year (days when the temperature falls between the 2.5th and 25th percentiles of the temperature distribution), while the impact of heat stems from high relative risks (RRs) occurring rarely (days when the temperature exceeds the 90th percentile of the temperature distribution). These results highlight the strong nonlinearity of the temperature-mortality relationship and the asymmetry between the temporal dynamics of the impacts of cold and heat. Repeating such studies over time will allow for testing hypotheses regarding spatiotemporal dynamics and, ultimately, will help inform the scale of surveillance and prevention efforts in the context of climate change.