BACKGROUND: The relationship between thermal stress and health has been only marginally investigated in North Africa. This study aimed to estimate the short-term effect of heat on total mortality, in the city of Tunis in 2005-2007, using time series analysis. METHODS: The study period was restricted to the summer season (May-October) and heat effect was assessed using maximum temperature as exposure variable. We estimated the breakpoint above which heat-related mortality begins to increase using a segmented linear regression. A Poisson Generalized Estimating Equations (GEE) model was then used to estimate the impact of heat on daily mortality. Models were adjusted for nitrogen dioxide (NO2), trend, calendar month, day of the week, the Ramadan period, and holidays. RESULTS: The estimated breakpoint was 31.5 degrees C (standard deviation: 0.9 degrees C). After adjustment for potential confounders, the daily mortality increased significantly by 2.00% [95% confidence interval: 0.68-3.16] for a 1 degrees C increase in daily maximum temperature above the breakpoint. An increase of 10mg/m(3) in NO2 was associated with a significant increase in daily mortality (0.48% [0.08-0.88]). CONCLUSION: There is an important effect of heat on daily mortality in the city of Tunis. This is the first evaluation of such an association in a North African city with hot and dry summers and a lower middle economy.