SummarySix cycles of reciprocal recurrent selection (RRS) between Line 8 (Duroc) and Line 9 (Beltsville No. 1) were completed at the Fort Reno Livestock Research Center, E1 Reno, Oklahoma. A randomly mated control line was maintained. Each cycle of selection required three seasons. Selection of purebred pigs (born in the second season) was based on the mean 21-d litter weight of their maternal and paternal half-sib crossbred gilts that were born in the first season and farrowed in the third season (XB21DLWT). On the average, 5.8 maternal and paternal half-sib gilts contributed to the mean XB21DLWT for each purebred individual. The average potential selection differential for XB21DLWT was 5.64 kg, but only 70.2% of this value was realized in the initial selection (3.95 kg). Disease problems and unsoundness were the primary reasons for this discrepancy. The potential, initial, final and weighted final standardized selection differentials for females averaged 61, 60, 55 and 47%, respectively, of the corresponding differentials for males. Standardized selection differentials were similar for Line 8 and Line 9. The estimate of realized heritability for XB21DLWT was .076 + .319 for the average of 8 • 9 and 9 x 8 gilts. Environ- Based on these results and considering the complexity of RRS, the increase in generation interval and the expected decline in purebred performance that theoretically should accompany successful RRS, it seems doubtful that RRS can be more beneficial than selection based on purebred performance as a method to improve productivity of crossbred gilts.