Genetic control of temperature tolerance was investigated in pearl millet at seedling stage. Ten parents were selected through a field screening technique. They were crossed in a diallel manner. Fortyfive F(1) hybrids along with their 10 parents were tested for performance in randomized block design with three replications under supra-optimal temperature exposure at seedling stage. The experiment was conducted in three environments at different dates of sowing from May to July 2005. The Griffings method of diallel analysis was used to obtain the genetic estimates. Variances due to general combining ability and specific combining ability for all the characters studied, viz seedling thermo-tolerance index, seed to seedling thermo-tolerance index, germination, emergence rate, leaves/seedling, seedling height, seedling fresh weight, seedling dry weight were highly significant in all the three environments (two stress and one normal) indicating the importance of both additive and non-additive genetic variances. The non-additive (dominance) component was more prominent for all the characters. It suggested that heterosis breeding could be successful. The genotype 'CVJ 2-5-3-1-3' and '(77/371xBSECT CP 1)' were identified as the best general combiners for both the heat tolerance indices. Seven hybrds, namely, 'H77/833-2x96AC-93', '1305x96AC-93', '(77/371xBSECT CP-1)xTogo II', 'H77/29-2xCVJ 2-5-3-1-3', '1305x99HS-18', 'G73 107x77/245' and 'H77/833-2xH77/29-2' were identified having high values of heat tolerance indices seedling thermo-tolerance index and seed to seedling thermo-tolerance index. The estimates of additive genetic variance and narrow sense heritability for seedling thermo-tolerance index and seed to seedling thermo-tolerance index were high.