Cat-scratch disease (CSD) is caused by Bartonella henselae, and possibly by B. clarridgeiae. In immuno-compromised persons, B. henselae is one of the agents causing bacillary angiomatosis. Domestic cats are the main reservoir of these bacteria, which are transmitted primarily from cat to cat by fleas. Possible strategies to prevent the spread of infection among cats are to eliminate flea infestation or to prophylactically immunize cats. In order to develop an appropriate vaccine, it is important to determine if cats become resistant to re-infection by the same strain or various types or species of Bartonella. In a series of experiments, 21 SPF cats were experimentally infected by the intradermal route with 10(5)-10(10) colony-forming units/ml of either B. henselae type II (17 cats), or a new strain 'Humboldt' isolated from a mountain lion (4 cats). The cats were bled weekly to every other week for determination of bacteremia and specific antibody production. After they cleared their infection, they were challenged by a homologous or heterologous strain of Bartonella: 10 cats were challenged with B. henselae type II, three cats with B. henselae type I, four cats with B. clarridgeiae and four cats with the 'Humboldt' strain. Seven of these cats received a third inoculum dose resulting in three cats sequentially infected with sequence B. henselae type II/B. henselae type II/'Humboldt', two cats with sequence B. henselae type II/'Humboldt'/B. clarridgeiae, and two cats with the sequence 'Humboldt'/B. henselae type II/'Humboldt'. All cats challenged with a homologous strain remained abacteremic after challenge and had an increased IgG antibody titer. All cats challenged with either a different Bartonella species or type became bacteremic. The few cats receiving a third inoculum with a strain homologous to the initial strain remained abacteremicafter that challenge. All cats infected with B. clarridgeiae suffered relapsing bacteremia compared to only 36% of the B. henselae infected cats and 22% of the 'Humboldt'-infected cats (p=0.008). The duration of bacteremia was significantly longer in B. henselae primary-infected cats (mean: 34 weeks) than B. henselae heterologously challenged cats (mean: 9 weeks) (p=0.014). These data clearly indicate the lack of cross-protection between B. henselae and B. clarridgeiae and furthermore, indicate the lack of protection between B. henselae types I and II, and a wildlife isolate. A vaccine strategy for CSD prevention in domestic cats will require a multivalent vaccine approach.