The aim of this double-blind pilot trial was to compare clindamycin/primaquine with trimethoprim-sulfamethoxazole (TMP-SMZ) as primary treatment for AIDS-related Pneumocystis carinii pneumonia (PCP). The focus was on toxicity and tolerability since comparisons of efficacy were limited by the small sample size. Sixty-five individuals with a first episode of possible PCP were randomly assigned to receive clindamycin/primaquine (34 patients) or TMP-SMZ (31 patients). PCP was subsequently proven microbiologically in 27 and 22 of the patients in these respective groups. Half of the participants had an arterial partial oxygen pressure at enrollment of < or = torr. The incidence and severity of adverse reactions were lower--but not significantly lower (P = .07 and .08, respectively)--with clindamycin/primaquine. The markers of severity improved in a similar manner regardless of which regimen was administered. No significant differences were documented in outcome, duration of survival, length of the PCP-free interval, or rate of relapse. The results of this pilot study show a trend toward less toxicity with clindamycin/primaquine than with TMP-SMZ. This result must be confirmed by larger-scale clinical trials, which are also needed to better compare the efficacy of the two regimens.