In Purple Martins Progne subis, adult males were found to have significantly longer wings and tails than subadult males, whereas females of both age classes were equal in flight feather length. This male age class difference was not caused by higher mortality of long-winged subadult males because the pattern also existed for the same individuals measured at age one and two years, indicating that males postponed full feather growth for their first breeding year while females did not. Another male age class difference was that adult males were paired to their mates assortatively by wing-length while subadult males were not. This difference suggests either that females paired to subadult males were constrained in their choice of males or they did not prefer to pair with longer-winged males. A possible lack of female preference for longer-winged subadult males is consistent with two apparent costs that females suffered by pairing with subadult males that had longer wings than themselves: (1) The greater the difference in wing-length between subadult males and their mates, the less the male provisioned the brood relative to the female, and (2) in a parallel study, females paired with longer-winged subadult males obtained many fewer extra-pair fertilizations from adult males. These findings suggest that, contrary to previous hypotheses, ''delayed size maturation'' may be primarily sexually selected in some species, rather than naturally selected.