Female mites of the genus Varroa reproduce on the immature stages of Apis cerana F. and A. mellifera L. Mites are found more often in drone brood than worker brood, and while evolutionary explanations for this bias are well supported, the proximate mechanisms are not known. In one experiment, we verified that the proportion of hosts with one or more mites (MPV, mite prevalence value) was significantly greater for drones (0.763 ± 0.043) (lsmean ± SE) than for workers (0.253 ± 0.043) in populations of mites and bees in the United States. Similar results were found for the average number of mites per host. In a second experiment, using a cross-fostering technique in which worker and drone larvae were reared in both worker and drone cells, we found that cell type, larval sex, colony and all interactions affected the level of mites on a host. Mite prevalence values were greatest in drone larvae reared in drone cells (0.907 ± 0.025), followed by drone larvae reared in worker cells (0.751 ± 0.025), worker larvae reared in worker cells (0.499 ± 0.025), and worker larvae reared in drone cells (0.383 ± 0.025). Similar results were found for the average number of mites per host. Our data show that mite levels are affected by environmental factors (cell type), by factors intrinsic to the host (sex), and by interactions between these factors. In addition, colony-to-colony variation is important to the expression of intrinsic and environmental factors.