Selection was practised in two replicates for both high and low testis weight in the mouse. Typically 7 males were selected out of 30 recorded for a total of 5 generations. From an initial average of 191 mg the mean divergence between high and low lines reached 112 mg, with a realized heritability of 0·52. The ovulation rate of the lines changed in the same direction as that of selection, the mean divergence was 2·0 eggs in primiparous females in generation 4 and 1·6 in nulliparous females in generation 5. Correlated changes in the body weight of both sexes also occurred but were inadequate to account for the observed change in ovulation rate. The genetic regressions of ovulation rate on testis weight were estimated to be 2·9 and 14 eggs/100 mg in primiparous and nulliparous females, respectively, which, along with data from other experiments, correspond to genetic correlations between testis weight and ovulation rate of 0·50 and 0·25 respectively. There were no correlated changes in litter size. The possibility of using male testis size in breeding programmes to improve female reproductive performance is discussed.