Females of the parasitoid wasp Goniozus nephantidis paralyse host caterpillars and lay a clutch of up to 18 eggs onto the host integument. The known biology of G. nephantidis suggests that matings occur exclusively between siblings from the same brood. This leads to the prediction that brood sex ratios should be highly female-biased and have low variance. Sex ratios are indeed female-biased, with the mean proportion of males equal to 0.093. However, while sex ratio variance is significantly less than binomial, many broods contain no males at emergence. During development 28% of G. nephantidis offspring die. Male mortality offers a potential explanation for all-female (= virgin) broods. For the clutch sizes and mortality observed, theory predicts that <10% of females will emerge from all-female broods but the empirical value is much higher. The prediction that the prevalence of virginity decreases with increasing clutch size is, however, supported. We consider alternative explanations for the observed proportion of all-female broods, but this appears to be neither an artefact of the laboratory environment nor due to incorrect assumptions about G. nephantidis life history. Although its reproductive biology has been much investigated and its sex ratio matches some theoretical predictions, we conclude that a fuller understanding of G. nephantidis sex ratio requires a deeper knowledge of its field biology.