The single- and multiple-dose pharmacokinetics of and tolerability to a new microfluidized suspension of atovaquone were studied in human immunodeficiency virus-seropositive patients with CD4 counts of < or = 200 cells per mm3 in order to define a dosing regimen for the treatment of Pneumocystis carinii pneumonia. This was an open study with groups of six patients each. In the first part of the study, six subjects received escalating single doses of 500, 1,000, and 1,500 mg after an overnight fast at weekly intervals. In the second part of the study, groups of six subjects were dosed for 14 days according to three regimens: 1,000 mg twice daily fasting, twice daily with a high-fat meal, or once daily with a high-fat meal. Plasma atovaquone levels were assayed by high-performance liquid chromatography. Pharmacokinetic parameters were determined by noncompartmental methods, and statistical comparison of parameters for single doses was performed by analysis of variance. Plasma drug concentrations increased with single doses from 500 to 1,000 mg but were no higher with a dose of 1,500 mg. Thus, 1,000 mg was selected for multiple administration. A regimen of 1,000 mg twice daily with food resulted in a 93% increase in the average trough steady-state concentration compared with 1,000 mg once daily with food. Food increased the bioavailability of atovaquone 1.4-fold over that in the fasting state. All patients who received 1,000 mg twice daily with food achieved target steady-state concentrations in plasma of 15 to 25 micrograms/ml. Multiple-dose regimens were generally well tolerated, but the higher levels in plasma achieved by 1,000 mg twice daily with food were associated with an increased incidence of rash. In conclusion, target plasma atovaquone concentrations for the treatment of P. carinii pneumonia can be achieved in most patients with 1,000 mg twice daily in a fasting state and in all patients with 1,000 mg twice daily administered with food, but at higher concentrations in plasma, there may be an increased risk of rash.