Although important in epidemiological theory, the relationship between the size of host populations and the prevalence of parasites has not been investigated empirically. Commonly used models suggest no relationship, but this prediction is sensitive to assumptions about parasite transmission. In laboratory populations, I manipulated the size of Tribolium castaneum flour beetle populations and measured the prevalence and distribution of a parasitic mite, Acarophenax tribolii. I found that parasite prevalence did not vary for a wide range of host population sizes. However, prevalence was lower in populations with less than 40 hosts. This effect cannot be attributed to changes in host population density because host density was held constant among treatments. The reduction in prevalence of small populations below a threshold that I observed is predicted by the extinction debt model, but it is not expected from models of host–parasite interactions that assume density-dependent transmission. The distribution of parasites, measured using Lloyd's patchiness index, was not affected by host population size. The mean crowding of parasites, however, was negatively related with host density. Finally, the prevalence of parasites in large populations did not differ from that found in sets of smaller patches as long as the smaller populations in aggregate were equivalent in size to the large population.