Ecological specialization is hypothesized to result from the exploitation of predictable resource bases. For parasitic organisms, one prediction is that parasites of large-bodied host species, which tend to be long-lived, should specialize on these hosts, whereas parasites of small host species, which represent more ephemeral and less predictable resources, should become generalists. We tested this prediction by quantifying the association between the level of host specificity of fleas and the mean body mass of their mammalian hosts, using published data from 2 large, distinct geographical regions (South Africa and northern North America). In general, we found supporting evidence that flea host specificity, measured either as the number of host species exploited or their taxonomic distinctness, became more pronounced with increasing host body mass. There were, however, some discrepancies among the results depending on the different measures of host specificity, the geographical region studied, or whether we used the raw values or phylogenetically independent contrasts. These are discussed with respect to other forces acting on the evolution of host specificity in parasites, as well as in the context of the regions' contrasting evolutionary histories. Overall, though, our findings indicate that the exploitation of large-bodied, and therefore long-lived, host species has promoted specialization in fleas, most likely because these hosts represent predictable resources.