Reciprocal recurrent selection in the maize (Zea mays L.) populations, BSSS(R) and BSCB1(R), was compared with recurrent selection for general combining ability in BS13(S(,2)), and half-sib selection in the intervarietal cross, BS6(RS). The original and improved populations and population crosses, and crosses among selected lines derived from the improved populations were grown at four locations in 1980 and 1981. The populations and population crosses were also grown as a separate experiment at Ames both years for more extensive agronomic evaluation;Selection significantly improved the yield of BSSS(R) and BS13(S(,2)); however, genetic drift may have caused significant yield reductions in BSCB1(R) and BS6(RS). BSSS(R) x BSCB1(R) and BS13(S(,2)) x BSCB1 (R) were improved similarly for yield indicating that both types of selection were equally effective for improving the Stiff Stalk Synthetic x Corn Borer Synthetic cross. Population crosses involving BS6(RS)C2 were significantly lower yielding than the other crosses. Several yield components were improved with selection for yield;Four selected S(,2) lines derived from the improved populations were crossed in a design II mating to evaluate their performance and type of gene action. The best crosses from each set did not yield significantly more than the population crosses but did compare favorably to the check hybrids for overall agronomic performance. For all traits, most of the variation was attributed to general combining ability of the lines; consequently, selection seems to have been primarily for favorable alleles at loci exhibiting partial to complete dominance;Genetic effects of selected lines derived from BSSSCO and BS13(S(,2))C1 were compared using a diallel mating scheme within both groups of lines. Nonadditive genetic variance was greater than additive variance for the BSSSCO lines; however, additive genetic variance was more important for the BS13(S(,2))C1 lines.