Genetic factors affecting female reproductive performance in lines of mice with a known history of selection were estimated from a 5 X 5 diallel cross. Lines were selected as follows: large litter size at birth (L+); large 6-wk body weight (W+); an index for large litter size and small 6-wk body weight (L+W-); the complementary index (L-W+) and randomly (K). Partitioning of direct and correlated responses for litter size, 6-wk body weight and related traits into average direct genetic (li) and average maternal genetic (mi) effects indicated that the magnitude of differences in li exceeded those in mi. Lines having positive responses in li were W+ greater than L+ greater than L-W+ for dam body weight, L+ greater than L+W- greater than W+ for litter size and L+ greater than (W+, L+W-) for litter birth weight, whereas L-W+ responded negatively for litter size. A positive association was found between mi for litter size and dam body weight, W+ and L-W+ being high and L+ and L+W- low for both traits. Female infertility and time from male exposure to parturition had relatively small correlated responses. Line rankings in general combining ability (gi) and net line effects were similar for the respective traits. Depending upon the line and trait involved, the relative contribution of average direct genetic and line direct heterotic (hi) effects to general combining ability [gi = (1/2) li + hi] varied. Line heterosis refers to average heterosis in crosses involving that line. Direct heterosis ( hij ) for each trait differed considerably among crosses. The three crosses showing the highest hij for litter size at birth, W+ X L-W+ (1.78), L+ X W+ (1.28) and L-W+ X L+W- (1.22), possibly had loci contributing directional dominance to litter size with frequencies of parental lines deviating in opposite directions relative to mean gene frequency. The correlation between absolute difference in parental line means and hij for litter size was not significant, suggesting that the magnitudes of absolute differences in parental means were not reliable predictors of divergence in gene frequency. Crossbred performance increased linearly with midparent values for litter size at birth (b = .88 +/- .09, R2 = .92) and dam parturition body weight (b = 1.13 +/- .04, R2 = .99), the latter trait showing an increase (P less than .01) in heterosis as midparent values increased.