The objective of this study was to determine whether offspring sex ratios in Mew Gulls (Larus canus) change non-randomly according to the current condition of females, empirically evaluating the predictions of the cost of reproduction hypothesis in colonies on the islands of the middle reaches of the Vistula River in Poland. It was assumed that Mew Gull females in poor condition would produce more daughters (10-15% smaller than males). The condition of females was reduced by removing first clutches shortly after completion, inducing a second clutch. Non-manipulated pairs that laid first clutches at the same time constituted controls. In first clutches (control, 3 eggs), the mean proportion of daughters at hatching was 51.0 ± 15.4% (n = 50), and on the fifth day after hatching was 51.2 ± 12.0% (n = 48). The share of broods with a predominance of daughters in both periods was similar (range = 47.9-48.0%). In experimentally induced replacement clutches (3-egg), the mean proportion of daughters was higher than in first clutches both at hatching (69.3 ± 14.9%; n = 45) and 5 daysays after hatching (62.1 ± 16.2 %; n = 42). Offspring sex ratio at hatching reflected the current condition of females, corresponding with predictions of the cost of reproduction hypothesis. Frequency of replacement broods with a predominance of daughters declined between hatching (77.8%) and the fifth day after hatching (59.5%), reducing broods in favor of males; and more frequent losses were observed at 5 daysays after hatching in replacement broods (of which only ∼13% of broods were without losses) compared to first broods (46.0% without losses).