Sex allocation theory predicts that females should bias the production of offspring towards the sex that will maximize maternal fitness. Here we demonstrate evidence for nonrandom sex allocation by female ruffs (Philomachus pugnax), at both the individual and population level in relation to female condition. At the population level, female condition varies significantly across 3 years and is mirrored by population sex ratio, such that in years when females are in poor condition the population offspring sex ratio is female-biased, while in years when females are in better condition there was little or no bias. In the year when females were in overall poor condition, females in better condition produced more daughters. The same relationship is also revealed by comparing the sex ratios of individual females breeding in two consecutive years in different condition. As the condition of an individual female improves (across years) she tends to produce more female offspring. Although we have shown that, as in other birds, female condition is an important determinant of sex allocation, our results also suggest that such nonrandom allocation does not occur in every year, being particularly strong in a year when females, on average, are in poorer condition. We suggest that our results are consistent with the idea that skewing the sex ratio is likely to carry a cost to females and that it is adaptive only when the fitness differential between sons and daughters is sufficient to outweigh probable costs.