Several non-mutually exclusive hypotheses predict adaptive variation in the offspring sex ratio. When conditions for breeding are adverse, parents are predicted to produce more offspring of the less costly sex to rear (‘the cost-of-reproduction hypothesis’). Moreover, they also should produce the more dispersing sex in order to diminish future competition (‘the local-resource-competition hypothesis’). Here, we analyse brood sex ratio according to rearing conditions in the southern shrike Lanius meridionalis, a species with moderately reversed sexual dimorphism. Our results suggest that females are more costly to rear than males in this species. Adult females proved heavier than males, and female nestling tended to be heavier than male nestlings. Moreover, the greater brood reduction, the more male-biased was the brood, suggesting that brood reduction implied higher mortality in female nestlings. Consistent with these findings, the brood sex ratio was biased to the less costly sex (males) when breeding conditions were adverse (bad years or low-quality male parents), supporting the cost-of-reproduction hypothesis. By contrast, these findings did not support the local-resource-competition hypothesis, which predicted female-biased brood sex ratio under adverse conditions. As a whole, our results support the idea that birds adaptively modulate sex ratio in order to minimize reproduction costs.