The canonical model of sex-chromosome evolution predicts that sex-antagonistic (SA) genes play an instrumental role in the arrest of XY recombination and ensuing Y chromosome degeneration. Although this model might account for the highly differentiated sex chromosomes of birds and mammals, it does not fit the situation of many lineages of fish, amphibians or nonavian reptiles, where sex chromosomes are maintained homomorphic through occasional XY recombination and/or high turnover rates. Such situations call for alternative explanatory frameworks. A crucial issue at stake is the effect of XY recombination on the dynamics of SA genes and deleterious mutations. Using individual-based simulations, we show that a complete arrest of XY recombination actually benefits females, not males. Male fitness is maximized at different XY recombination rates depending on SA selection, but never at zero XY recombination. This should consistently favour some level of XY recombination, which in turn generates a recombination load at sex-linked SA genes. Hill-Robertson interferences with deleterious mutations also impede the differentiation of sex-linked SA genes, to the point that males may actually fix feminized phenotypes when SA selection and XY recombination are low. We argue that sex chromosomes might not be a good localization for SA genes, and sex conflicts seem better solved through the differential expression of autosomal genes.