Alvarez et al. (1984) have demonstrated that the frequency-dependent fitnesses, for both total and sexual selection, reported by Bundgaard and Christiansen (1972)-hereafter quoted as BC Christiansen et al., 1977). Clearly, Christiansen turns round our arguments and, moreover, he overlooks some aspects on fitness estimation. We have never affirmed that the sexual estimator confounds sexual selection and of mating, but these two phenomena are confounded by B&C. Thus, on page 447, to derive the formulas for the sexual fitness estimators, B&C state: The sexual fitness values ... are those created by deviations in the adult mating from the one expected had the unmated adult population mated at random. It is obvious that B&C do not properly differentiate sexual selection, which involves frequency of mating, and of mating, which is not a selective agent and concerns choice of mate. This misunderstanding of the operation of selection at the level of mating behavior is really the origin for constructing an inappropriate sexual estimator. As a consequence of this erroneous definition of sexual selection, the sexual fitness estimator is comparing two sets of genotype frequencies which are not implicated in the action of sexual selection, and so it produces erroneous fitness estimates. Therefore, the claim made by Christiansen that the problems discussed by Prout (1965) and Christiansen et al. (1977), which deal with the estimation of total selection, are the causes of the incorrect estimation of sexual selection, does not make sense. In the analysis of the mating pattern from the B&C article they develop three kinds of sexual fitness estimators in order to analyze the components of mating behavior: female and male mating success (W,, and Ws,,, respectively) and joint mating success ( W/s). We have shown that whereas W,i and W,, correctly estimate female and male sexual selection, W5Q/8 produces spurious frequency-dependent fitness estimates, and that non-significant deviations of random mating are occurring in their experimental populations (Alvarez et al., 1984). Nothing of this is indicated in the B&C work. Thus it seems to us that Christiansen (1984) reinterprets the B&C article again when he affirms that in he structural analysis of the mating they do separate sexual selection from the combination rules in mating.