The purposes of this study were to select honey bees, Apis mellifera L., for resistance to varroa mites, Varroa jacobsoni Oudemans, and to find a probable cause for this resistance. As a genetic source, we assembled 8 colonies that we thought had potential for resistance to varroa. Queens and drones were propagated from this group to produce 43 instrumentally inseminated queens, each queen mated to only 1 drone. Colonies from 27 of these queens were tested in Louisiana and 16 were tested in Michigan. Each colony in the Louisiana test began with 986 ± 13 g (mean ± SD) of bees and ≈290 mites; Michigan colonies began with 3,212 ± 171 bees and ≈51 mites. The populations of mites and bees were measured 10 wk later. Three of the 43 colonies had fewer mites at the end of the test than at the beginning. During the experiment, we evaluated each colony for grooming behavior, hygienic behavior, the duration of the postcapping period, and the frequency of nonreproducing mites in brood cells. Of these 4 characteristics, only nonreproduction of mites was highly related to a change in the mite population. The duration of the postcapping period was marginally related, and the other 2 characteristics were apparently unrelated to the growth of the mite population. This study showed that resistance to varroa mites is present in the honey bee population in the United States, nonreproduction of mites was highly correlated with the growth of a mite population, and nonreproduction of mites may be a valuable characteristic for selecting bees for resistance to varroa mites.