Petrie et al. (1991) demonstrated a correlation between the degree of elaboration of peacocks' trains and their mating success, and also showed that this correlation occurred because females preferred to mate with the male that had the most elaborate train of those sampled on the lek. Although these data suggest that female choice is responsible for non-random mating in this species, they do not conclusively show that train morphology is the cue that females respond to, because they do not rule out the possible influence of another unidentified variable which is correlated with train elaboration. This paper presents an experimental test of the importance of the peacock's train in determining male mating success. If the number or arrangement of eye-spots in the peacock's train influences mating success, then changing the number of eye-spots should change mating success. This prediction was tested in an experiment where the trains of male peafowl (Pavo cristatus) were manipulated by removing a number of eye-spots between mating seasons. Peacocks with eye-spots removed showed a significant decline in mating success between seasons compared with a control group. This result, together with the observational data, supports the hypothesis that the peacock's train has evolved, at least in part, as a result of female choice.