Uterine capacity was estimated in Chinese Meishan (MS), Large White (LW) and crossbred (LW × MS) sows, using unilateral liysterectomy-ovariectomy (UHO) methods. In the first two parities, 20 sows of each genotype, whose left uterine horn and ovary had been removed in the post-pubertal period, were mated, their ovulation rate recorded by laparoscopy and allowed to farrow normally. In the third parity the mated sows were slaughtered at around 30 days of pregnancy and ovarian, uterine and embryo traits were recorded. The ovulation rate in UHO females was similar to that observed previously in intact females of the same population, MS sows producing 4·2 (s.e. 1·0) more ova at parities 1 and 2 and 7·0 (s.e. 1·9) more ova at parity 3 than LW sows. At parity 3, MS sows had 5·9 (s.e. 2·0) more viable embryos at 30 days of pregnancy than LW sows. Crossbred sows were intermediate for both ovulation rate and the number of viable embryos at 30 days. At the first two parities, litter size was higher in the MS by 1·0 (s.e. 0·8) piglets per litter than in the LW, while the crossbred sows exceeded the mid-breed mean by 1·7 (s.e. 0·7) piglets. Uterine capacity, estimated as twice the number of piglets carried to term in a single horn, was 12 foetuses in the LW, 14 in the MS and 16 in the crossbred sow. Piglets born to MS sows were 0·4 (s.e. 0·1) kg lighter than those born to LW sows, with a significantly lower (MS: LW = 0-68, P < 0·01) within-Utter standard deviation, which was found to be independent of the smaller mean weight. At parity 3, MS sows were found to have uteri of similar length with both lighter (-0·36 (s.e. 0·14) g) and shorter (-1·54 (s.e. 0·76) mm) embryos with smaller spaces (-7·8 (s.e. 2·7) cm) between embryo sites than LW sows. Within-Utter standard deviations for embryo size traits were lower in MS than in LW sows, especially for the distance between embryo sites (MS: LW = 0·45, P<0·01). The reduced variability of embryo traits in litters in MS sows was independent of their smaller litter means. Third parity litters from crossbred sows were similar to the mid-breed mean for means and variability of embryo traits but crossbred sows had significantly longer uteri (48·5 (s.e. 18·5) cm) than the mid-breed mean, being longer than either pure breed. It was concluded that differences between the pure breeds in uterine capacity may have arisen through better within-uterus organization, whereas the observed heterosis effects may result in part from physically longer uteri in crossbred sows and that uterine capacity contributes to the genotype differences in prenatal survival and prolificacy, especially in crossbred sows.