It is well known that weather conditions influence the comfort and human health, in particular temperature variations and its extremes. Many studies showed cold and heat-related mortality mainly in infants and the elderly. In Tunisia, the effects of temperatures have never been estimated at a large scale. The objective of this study was to estimate the effect of extreme daily temperatures on the risk of death in 7 different bioclimatic regions.

Generalized additive model was applied to assess the exposure-response relation and lag patterns of the association between mean temperature and the daily number of death from 1991 to 2011 in Tunis (North), Siliana and Jendouba (North West), Kairouan (Center West), Monastir (Center East), Gabes (South East) and Tozeur (South West). The analysis was adjusted for sex, age and population size, and controlled for long-term trend, seasonality and holidays. We did not adjust for the confounding effect of road traffic accident. The goodness-of-fit of the constructed model was assessed using generalized cross-validation (GCV) score and residual test.

Tunis registered 153,926 deaths, Siliana 20,937, Jendouba 36,974, Monastir 46,625, Kairouan 51,409, Gabes 34,347 and Tozeur 10,658. The coldest period was in January and February and the hottest one was in July and August for all regions. The higher mortality rate was observed in cold season. The effect of temperatures in mortality was immediate, and presenting a 1–26-days delay. The maximum risk of death was registered for under 10 degrees temperatures and for temperatures from 35 degrees and over.

Exposure to cold temperatures has more impact on mortality than hot weather. Climate in Tunisia is relatively warm; people are more sensitive to cold weaves. The health system should be prepared to reduce this impact mainly in vulnerable population such as children and old people.