Lentil (Lens culinaris Medikus) is a rain-fed crop in North Africa and West Asia whose yields are limited by the amount and distribution of rainfall and winter cold. This study aimed to quantify the effects of climatic variables on lentil seed yields through the fitting of simple empirical models to trial data of two cultivars sown at six sites, 1983-89, representing 31 environments in Syria. The ranges over environments were 152-527 mm for total season rainfall and 1-52 for the number of frost nights. Overall, the total seasonal rainfall accounted for 40.8% of the variance in mean seed yield (1.27 t/ha, S.D. 0.82) with a response of 5.68 kg/ha/mm. A multiple regression model with monthly rainfall from November to May explained 67.0% of the variance in mean seed yield. From November to February the response of seed yield to rain was <10 kg/ha/mm; rain in March, the period of late vegetative growth, made the most important contribution to seed yield. The response to April rain was negative. At Tel Hadya (the most frequently used site), the total seasonal rainfall accounted for 79.8% of the variance in mean seed yield, and the addition of the number of frost nights to the model improved the fit to 92.7%. Winter cold had a smaller effect on yield than rainfall, with no consistent overall effect, but differences over regions. The cultivars contrasted in their responses to drought (78S26002 was superior to ILL4400 at seasonal rainfall levels down to 134 mm) and the number of frost nights at Breda and Tel Hadya (78S26002 was more susceptible to cold than ILL4400). Thus, despite the predominant influence of rainfall on yield, the genetic variability in response to moisture and cold shows the scope for selection under rainfed Mediterranean environments.