Sex allocation theory predicts that parents should favor offspring of the sex that provides the greatest fitness return. Despite growing evidence suggesting that vertebrates are able to overcome the constraint of chromosomal sex determination, the general pattern remains equivocal, indicating a need for experimental investigations. We used an experimental feeding design to study sex allocation during 3 years in black-legged kittiwakes (Rissa tridactyla). Intense male-male competition for securing a breeding site is common in this species in which males are heavier and larger than females. Hence, we hypothesized that parents producing fledglings in better than average condition, as supplementarily fed pairs do, would increase their fitness return by producing sons. Conversely, producing daughters would be a better tactic for Unfed parents. Hence, we predicted that Fed parents produce more sons than Unfed parents. This prediction is particularly expected if sexual dimorphism arises as early as during chick rearing, suggesting strong selective pressures for optimal male development. Our results showed that 1) males were heavier and larger than females prior to fledging and that 2) Fed parents produced relatively more male hatchlings than Unfed parents. We interpret this result in terms of a Trivers-Willard-type process. Furthermore, our data revealed that Unfed parents significantly overproduced female hatchlings, whereas offspring sex ratio was balanced among Fed parents. Because the 3 reproductive seasons we considered were particularly poor food years, Unfed parents may have overproduced daughters to avoid the apparent higher reproductive costs of raising sons.