the objective of this trial was to evaluate the milk yield of cows of two breeding systems, straightbreds charolais (ch) and nellore (ne) and, crossbreds f1 ？ch ？ne and ？ne ？ch, submitted to different feeding systems, as well as measure the resultant heterosis. the milk yield was determined at 14, 42, 70 and 90 days, being this last one coincident with weaning. eighty-six cows were used, and submitted to one of the following feeding systems: np: cows maintained exclusively on native pasture (np); cp + np: cows maintained on cultivated pasture (cp) from july 15th until september 15th and the rest of the time on np; and np + cp: cows maintained on cp from september 15th until november 15th, and the rest of the time on np. cp was composed by oat (avena sativa), ryegrass (lollium multiflorum) and clover (trifolium vesiculosum). since no significant difference was verified between cp + np and np + cp feeding systems, they were combined in one feeding system (ncp). cows maintained on np showed lower average daily (3.93 vs. 5.46 l) and total milk yield (291.4 vs. 401.7 l), in relation to cows maintained on ncp, as well as displayed higher reduction of the milk yield (39 vs. 27%, respectively). average daily and total milk yield were not affected by cow genetic group within breeding system. f1 cows showed higher milk yield during the whole lactation period, being the heterosis for average daily milk yield more expressive for cows maintained on lower feeding conditions, exclusively on np (92.4%) in relation to cows with access to ncp (18.1%). the lactation persistency was higher for cows sired by ne bulls than for cows sired by ch bulls, and for straightbreds in relation to f1 cows.