Abstract Data on Brahman and Hereford cattle and their crosses were used to estimate direct and maternal additive and heterosis (dominance+epistasis) effects on cow productivity and calf preweaning growth traits. The cows, which were born in the subtropical environment at Grafton, Australia from 1983 to 1985, had varying levels of Brahman/Hereford genes. At the start of this study in 1990, half of the cows were transferred to the temperate environment at Ebor and the other half remained at Grafton. In both environments the cows were mated to Brahman (B) Hereford (H) and first-cross B×H bulls over three mating seasons. The same individual bulls were used in both environments. All the traits were influenced by a significant genotype by environment interaction, which was a cumulative effect of some or all of additive×environment, dominance×environment and epistasis×environment interactions. Weight of calf weaned per cow exposed to bull (used as an overall measure of cow productivity) was significantly influenced by a positive direct heterosis effect of 25.8% (relative to straightbred cows) and 33.2% at Grafton and Ebor, respectively. As traits of the calf, preweaning ADG and weaning weight were influenced by a significant negative Brahman direct additive effect at Ebor but not at Grafton. Direct heterosis effects were 10.8% and 10.3% at Grafton and 9.7% and 9.1% at Ebor for preweaning ADG and weaning weight, respectively. Inclusion of epistasis effects in the model significantly improved the accuracy of predicting calf performance of genotypes at Grafton. Therefore it is recommended that, where data structure allows, epistasis effects should be included in crossbreeding models.