The objective of this study was to determine the relative merits of four types of testers to evaluate maize (Zea mays L.) inbred lines for hybrid development and for reciprocal recurrent selection. From two brachytic (br2) maize populations, 100 S2 lines were derived and crossed to two broad-base testers (parental and opposite populations) and to two narrow-base testers (unrelated single-cross and inbred-line). In all instances, variance component estimates of lines were greater than their respective line x tester interactions. For grain yield, narrow-base testers had greater tester x location interaction than did broad-base testers. There was no trend for higher tester x line interaction in narrow-base testers than in broad-base testers. The genetic variance component among testcrosses, heritability coefficients, and percentage of coincidence of lines selected for general combining ability (GCA) were greater for narrow- than for broad-base testers when an unrelated line was used as tester. However, these estimates were lower for narrow- than for broad-base testers when the highest GCA inbred line was the tester. The parental-population tester had the most consistent discrimination of the variation among the S2 lines;The results suggest that inbred lines can be effectively used for identifying lines having good GCA. There was evidence that a line with high GCA was the poorest tester for estimating the variation among lines due to the masking effects of dominant alleles in the tester. Phenotypic and genotypic correlation coefficients between grain yield and ear height and yield and days-to-flower indicate that selection for improving grain yield will increase ear height and reduce days-to-flower. Selection for modifier genes for higher plant and ear height in br2 populations will help to overcome the negative effects of the br2 allele on grain yield;The estimates of the additive genetic variance were higher than those for normal maize populations. Relatively large genetic gains are expected in both populations, but no increase in the heterosis is expected with one cycle of reciprocal recurrent selection. The results suggested that recurrent selection methods can be effectively used to overcome the negative effects of the br2 gene on grain yield.