Half-sib and full-sib reciprocal recurrent selection (RRS) in maize ( Zea mays L.) have been successful methods of simultaneous, reciprocal improvement of two populations to maximize the performance of the population cross. Objective of this study was to use computer simulation to compare responses to selection of populations themselves and their crosses for half-sib RRS and full-sib RRS and their modification. The modification included an additional generation of inbreeding before producing testcross progenies (S 2 plants as recombination units). Parental populations included 110 individuals throughout selection cycles. Evaluation of each individual in both RRS methods was based on the mean performance of 10 testcross progeny. Five females per male were used to produce testcross progenies in half-sib RRS. Selection intensity was 10% in all instances.;There were 22 initial conditions corresponding to two sets of parental initial population gene frequencies and 11 genetic models. Two sets of parental initial populations gene frequencies were equal and unequal. The genetic models used were additive, dominance (3 models), and epistasis (7 models). Half-sib RRS and full-sib RRS and their modification were simulated for each of 22 initial conditions. Three replications of each initial condition were conducted for 20 cycles of selection.;The simulation study indicated that the efficiency of full-sib RRS was similar to half-sib RRS for 21 initial conditions with S 1 plants as recombination units. The linear response of half-sib RRS was 1.7 times significantly (p < 0.01) greater than full-sib RRS for the genetic model that included complete dominance and dominance x dominance epistasis with S1 plants as recombination units. There were no significant differences, however, between full-sib and half-sib RRS with S2 plants as recombination units for all 22 initial conditions. Use of S2 plants as recombination units increased selection response from both full-sib RRS and half-sib RRS. Compared with half-sib RRS, full-sib RRS evaluates 50% fewer testcross progenies, but has the same efficiency of half-sib RRS. Full-sib RRS with S2 plants as recombination units could be a more efficient method of interpopulation improvement in maize breeding.