The number of lines recombined in a recurrent selection program affects the genetic variation remaining after selection as well as the selection intensity. Both of these variables are important in establishing and maintaining a desirable rate of progress from recurrent selection. Little empirical evidence is available in maize (Zea mays L.) to establish the appropriate effective population size for recurrent selection programs. The objective of this study was to compare the response to Si-progeny selection in four selection programs (5-S1, 10-S1, 20-S1, and 30-S1). Each program had different numbers of individuals intermated (5,10, 20, or 30, respectively) with a common selection intensity of 20%. Four cycles of selection were completed for the 5-S1, 10-S1, and 30-S1 programs and five cycles of selection were completed for the 20-S1 program. Grain yield increased significantly for the 10-S1, 20-S1. and 30-S1 programs, and there were no significant differences in rates of response among programs. Grain yield decreased signficantly in the 5-S₁ program. All four programs improved significantly for grain yield of testcrosses with ‘Pioneer Two-ear Composite’, Cycle 0 (BS11CO) and the inbred B79. These results suggest that drift becomes a stronger force in altering allele frequencies than selection when fewer than 10 lines are recombined. There does not seem to be an advantage, at least in the short term, of recombining more than 10 lines per cycle of selection.