Male-lethal, maternally inherited spiroplasmas occur in four species of Drosophila, and persist in natural populations despite imperfect vertical transmission rates. In the field, larval crowding is thought to be sporadic, but occasionally intense. To determine whether crowding affects host persistence, I compared the population dynamics of infected females (hosts) under crowded conditions to those expected from data collected on uncrowded females. I estimated host fitness components and maternal transmission rates for individual females under uncrowded conditions in both the artificial host D. pseudoobscura (this paper) and the native host D. willistoni (previously reported). Spiroplasma infection had no effect on lifetime production of daughters in D. pseudoobscura; however, as with some D. willistoni lines, hosts may produce more of their daughters earlier in life than nonhosts. Because individual contributions to relative rates of increase calculated from these fitness data were similar for hosts and nonhosts, I expected hosts to persist in laboratory populations. Instead, three patterns were observed: rapid extinction of D. willistoni females infected with male-lethal spiroplasmas, slow decline or persistence of hosts (depending on initial frequency) in both D. pseudoobscura infected with male-lethal spiroplasmas, and D. willistoni infected with non-male-lethal spiroplasmas. Population dynamics, then, depend on host species and bacterial isolate. Fitness estimates change with host line in uncrowded D. willistoni, but host genetic background did not affect population dynamics. These and previously published results show that the interaction phenotype changes with host and parasite isolate, and that host fitness can be affected by crowding. Crowding in natural populations may therefore decrease host fitness but, in expanding populations, early reproduction in hosts may be to their advantage. Possible effects of seasonal fluctuations in population density on the fitness of infected Drosophila are discussed.