Aggression among spotted hyaena, Crocuta crocuta, siblings is often intense, and sometimes lethal. Frank et al. (1991, Science, 252, 702-704) proposed that siblicide routinely occurs in half of all spotted hyaena litters, namely those composed of same-sex twins. We propose an alternative to this 'obligate' model. In our 'facultative' model we suggest that siblicide is far less common than previously supposed, and that it occurs only when resources are insufficient to sustain two cubs. According to this facultative model, intense neonatal aggression functions to establish intralitter dominance rather than to kill siblings. Furthermore, differences in litter size and composition between captive and field settings previously used to support the obligate model are assumed in the facultative model to be due to prenatal factors rather than to postnatal siblicide. Here we tested the predictions of these two hypotheses with 10 years of field data from hyaenas inhabiting the Masai Mara National Reserve, Kenya. We found that, although sex compositions of hyaena litters averaged over the entire 10-year period did not differ from those predicted by chance, they did vary with environmental conditions. Litter sizes, by contrast, remained constant. These data are inconsistent with the widely accepted hypothesis that spotted hyaenas in same-sex litters routinely engage in siblicide. Copyright 1999 The Association for the Study of Animal Behaviour.