Geographical variation in forest insect sensitivity to environmental conditions is often overlooked. Yet considering regional variation in response to climate is necessary to better predict the consequences of climate change on pest population dynamics and associated damage. We investigated the climatic drivers of demographic fluctuations in a forest tree defoliator, the pine pro cessionary moth (PPM), in 8 biogeographical regions in France. We studied the effects of precipitation and temperature for both cold and warm periods, in accordance with PPM biology. We trained second-order log linear models of infestation dynamics based on records spanning over 32 years (1981-2014) and 1239 forest plots, and tested their forecast accuracy on two additional years (2013-2014). PPM population growth rate decreased with higher precipitations in 5 regions and increased with increasing temperatures in the cold period in 4 regions. The magnitude and direction of temperature effects in the warm period varied among biogeographical regions. Our results also revealed that PPM population dynamics were largely triggered by density-dependent factors independent from climatic drivers. Short-term forecasts were generally accurate irrespective of the inclusion of climatic variables into the models. To better predict the consequences of climate change on insect population dynamics, it is advisable to take geographical variability of species response into account and develop regional forecasting models.< Réduire