In order to investigate the proposal that modifications in developmental rate may mediate differences in aggressive behavior, two series of selective breeding studies were completed. Stable lines of mice that differed in the frequency and latency of attacks were rapidly established in two series (by S1 in the first series and S3 or S4 in the second series). For evaluation of the developmental-genetic proposal, an analysis was made of the ontogeny of aggressive expression in male mice of the two series, with a detailed report provided of the S1 and S4 generations of the second breeding series. Comparisons between the results of a longitudinal design and the results of a new type of cross-sectional design (involving only siblings, or co-sibial) indicated (a) a reliable developmental course of attack expression, with a sharp rise in early maturity and a slow decline thereafter; (b) a convergence in later maturity of the behavior of lines selectively bred for high or low aggressive behavior, if the animals had been assigned to the longitudinal design; (c) a strong effect of repeated testing on attack latency and frequency, even though the dyadic tests were brief in duration and separated by long intervals. In addition, cross-generational comparisons suggested that the selective breeding differences came about primarily by changes in the behavior of the low aggressive lines, in that these animals failed to show in early maturity the sharp increases in attack occurrence that were observed in earlier generations. Certain implications of these findings for developmental and evolutionary concepts (e.g., neoteny, acceleration, heterochrony) are discussed.