Delineation of the selective pressures responsible for the evolution of sterile worker castes found in social insect colonies remains a major unsolved problem in evolutionary biology. There has therefore been a great deal of interest in suggesting ways by which the inclusive fitness of sterile workers can potentially be larger than those of solitary nest-builders. Queller's (1989) head-start hypothesis (Proc. natn. Acad. Sci. U.S.A. 86, 3224) suggests that workers may gain relatively more inclusive fitness because they have access to young of various ages which can be quickly brought to the age of independence, whereas a solitary foundress has to survive for the entire duration of the development of her brood. I argue here that Queller's quantitative analysis is incorrect because it gives an unfair advantage to workers, either by giving full credit of rearing an offspring to a worker who only cared for it for a short while or, by assuming that a worker can do much more work per unit time than a solitary foundress. I show, however, that workers do indeed have an advantage over solitary foundresses because they have assured fitness returns, even if in small amounts, for short periods of work. This results from a different reckoning from that used by Queller and gives a more moderate advantage, arising essentially from saving the wasted effort that occurs when lone foundress nests fail. Using field and laboratory data on the primitively eusocial wasp Ropalidia marginata, and re-analysing data on the four species of polistine wasps used by Queller, I show that such an `assured fitness returns' model provides a selective pressure for the evolution of worker behaviour which is at least about as strong as that of haplodiploidy, but free from such requirements of the latter as high levels of worker-brood genetic relatedness and ability of workers to manipulate brood sex ratios.