The combination of primaquine with clindamycin is effective in both in vitro and in vivo models of Pneumocystis infection. Primaquine alone at concentrations from 10 to 300 ,ug/ml reduced the numbers of organisms in cultures to less than 7% of control. Significant inhibition was observed down to 0.1 ,ug/ml. Clindamycin at 5 ,ug/ml was ineffective alone. Combinations of clindamycin and primaquine in culture at various concentrations were effective, but there was no evidence of true synergy. In rats with established Pneumocystis pneumonia, clindamycin alone at 5 or 225 mg/kg was ineffective. Primaquine alone at 0.5 or 2 mg/kg did not significantly affect the numbers of organisms remaining. The combination of 0.5 mg of primaquine per kg and 225 mg of clindamycin per kg was effective for therapy, lowering the numbers of organisms in the lungs by about 90%. The combination of 2 mg of primaquine per kg and 225 mg of clindamycin per kg was more effective, lowering the numbers of organisms by almost 98%. In the in vivo prophylaxis model, primaquine at 0.1 or 0.2 mg/kg did not prevent the development of Pneumocystis pneumonia in immune-suppressed rats. Clindamycin at 50 mg/kg had a modest effect alone, but at 5 mg/kg all animals became heavily infected. At 0.5 mg/kg, primaquine alone reduced the severity of infection, but seven of eight rats were still infected. In contrast, the combination of 5 mg of clindamycin per kg and 0.5 mg of primaquine per kg prevented infection in 8 of 10 rats; 2 rats had minimal infection. These studies suggest that the combination of clindamycin and primaquine should be tested in therapy or prophylaxis of Pneumocystis infections in humans.Treatment or prophylaxis of pneumonia caused by Pneumocystis carinii in patients with acquired immune deficiency syndrome (AIDS) has been hampered by the lack of effective, nontoxic agents. Pentamidine causes a variety of side effects which may limit therapy, and relapses are frequent (29,34). Trimethoprim with sulfamethoxazole causes a surprisingly high incidence of side effects in AIDS patients, including severe hypersensitivity reactions that limit use for treatment (29,33,34). Other experimental agents have been tested with variable results. Difluoromethylornithine is effective in some patients, but thrombocytopenia may be a limiting toxicity (9, 34). Pyrimethamine with sulfadoxine (Fansidar) or dapsone with trimethoprim may be effective in individual patients, but significant adverse reactions still occur with both combinations (10, 21, 34). Trimetrexate, a lipid-soluble analog of methotrexate, is being investigated (1).In the course of screening antimicrobial agents for potential effectiveness against P. carinii, we have evaluated combinations as well as individual agents. One of the combinations we chose to test was clindamycin with primaquine. This selection was based on the observation by Schmidt (25) that the curative action of primaquine in malaria was enhanced by mirincamycin. Mirincamycin, like clindamycin, is a lincosamide but has not been exte...