Twenty-eight F1 hybrids of wheat and their parents were grown in field trials at Trumpington, Cambridge during 1986/87 and 1987/88. They were derived from crosses between seven ‘modern’ varieties, used as female parents, and either two ‘old’ (Squareheads Master and Partridge) or two ‘modern’ varieties (Bert and Motto), which were used as male parents. Grain yield, yield components, biomass and height were determined. The male parents were chosen to provide contrasting phenotypes and genetic backgrounds for the F1 hybrids. Mid-parent advantage, the increase of a hybrid for a given character above the mean of its parents, and heterosis, the increase of a hybrid above the ‘better’ parent for that character, were calculated. Most F1 hybrids showed mid-parent advantage for the characters studied. This tended to be greatest for hybrids derived from parents with the largest phenotypic differences in that character. In contrast, where heterosis occurred it tended to be greatest where the phenotypic difference between the parents was least. This suggests that the beneficial effects of hybridization, resulting from the dispersion of dominant genes between the parents, was insufficient to overcome the detrimental effects of other genes present where the ‘less good’ parent was substantially lower than the ‘better’ parent. Hybrids derived from the ‘modern’ male parents had greater heterosis for grain yield and mean grain weight than those from the ‘old’ parents. Of the yield components, positive heterosis for mean grain weight resulted in heavier seeds and was the most important yield component in determining heterosis in grain yield. Heterosis for the number of grains/ear was small or did not differ significantly from zero while number of ears/m2 showed negative heterosis resulting in fewer ears/m2 in the hybrids.