It has been argued recently that the combination of male–only parental care and classical polyandry in birds is the most interesting and yet the least understood of all avian breeding systems. Despite a huge number of hypotheses, careful comparative analyses have repeatedly failed to identify consistent ecological differences between species showing male–only care and closely related species showing other patterns of care. This has led to the suggestion that such analyses fail because the crucial differences are between ancient lineages rather than between closely related species. Here, therefore, I use comparisons between families to test three well–known hypotheses: that male–only care is associated with: (i) a low rate of fecundity; (ii) large egg size relative to female size; or (iii) female–biased opportunities for remating. Families showing male–only care do not differ from families showing female–only care with respect to rate of fecundity or relative egg size. There is, however, a significant difference between these two groups of families with respect to an index of remating opportunities, nesting density. Families showing female–only care nest at high density, while those showing male–only care nest at very low density. This is one of the first times a consistent ecological correlate has been identified for male–only care in birds. It suggests that female–only care arises (or persists) in families where remating opportunities are abundant for both sexes, whereas male–only care arises (or persists) in families where remating opportunities are rare for both sexes and particularly scarce for males. This in turn suggests that sex differences in remating opportunities are the key ecological factor in determining male–only care and classical polyandry in birds.