Eleven cycles of reciprocal recurrent selection (RRS) were conducted in two maize (Zea mays L.) populations, BSSS(R) and BSCB1(R). The objectives were to evaluate the direct and indirect response to selection, combining ability of populations with related and unrelated inbred testers, and effects due to selection and genetic drift on response to selection.;The direct response for grain yield observed in the population crosses was greater than indirect responses observed in the populations per se. The observed rate of response for grain yield was significant in BSSS(R) x BSCB1(R) and BSCB1(R) but a significant response was not observed in the BSSS(R). Changes in other agronomic traits including grain moisture, root and stalk lodging, ear and plant height, and silking and pollen date were generally in desired directions.;The genetic model indicated that selection increased the frequency of favorable alleles affecting grain yield in the improved populations of BSSS(R) and BSCB1(R). The lack of improvement for grain yield observed in both populations was attributed to the effects of inbreeding due to genetic drift. The estimated indirect responses for grain yield, adjusted for effects of genetic drift, were similar to the estimated direct response.;The increased heterosis for grain yield in the interpopulation crosses with cycles of selection was interpreted as resulting from selection for alleles at complementary loci in each population and fixation of different alleles at a locus in each population by genetic drift.;RRS was effective in improving general as well as specific combining ability of the populations, as suggested by the increases in grain yield of the interpopulation crosses and the testcrosses of populations with related and unrelated inbred testers.;The results of this study suggest that RRS is an effective method of selection for improving the population cross. The improved populations can be used as potential germplasm sources in an applied breeding program.