Detailed observations on the acquisition and propagation of experimental Rickettsia mooseri infection in two species of fleas are presented. Rickettsia mooseri infection became detectable by means of the direct fluorescent antibody test about 2 days earlier in Leptopsylla segnis than in the putative vector, Xenopsylla cheopis. By the 6th day after the infective feeding, the entire lining and the lumen of the midgut in L. segnis contained masses of rickettsiae and the agent was being passed in the feces of the flea, whereas in X. cheopis these events did not occur until the 8th day. Basic behavioral differences in the two species of flea may explain these discrepancies and also influence their ability to serve as vectors of murine typhus. As a semisessile flea and sustained feeder, L. segnis only rarely attaches to a second individual and thus has an opportunity to acquire a heavy dose of rickettsiae, if feeding on a rickettsemic host. X. cheopis, in contrast, feeds rapidly and intermittently, even on man, and generally leaves its host soon thereafter, later returning to the same or another host to feed again. While L. segnis may not be as efficient a vector as X. cheopis regarding the intramurine cycle or transmission to man of murine typhus, the dense accumulation of infective feces on certain sites on the fur of the host raises the possibility of air-borne infection to man or rodent. Infection with R. mooseri had no effect on the survival of X. cheopis and L. segnis. Furthermore, no visible cytopathological effect was found in the paraffin-embedded sections of infected fleas.(ABSTRACT TRUNCATED AT 250 WORDS)