One hundred sixty patients were entered in two multicenter protocols to receive 400 to 2,000 mg of ketoconazole once daily for nonmeningeal or meningeal coccidioidomycosis. For 24 h after administration of all doses, mean concentrations in serum exceeded MICs for Coccidioides immitis (trough concentrations, >1 μg/ml). Mean peak concentrations occurred 4 to 6 h after administration, ranging from 7 to 17 μg/ml for doses of 400 to 2,000 mg. Incremental increases in peak concentrations in serum were greatest at doses of ≤1,200 mg. To investigate whether long-term therapy altered concentrations in serum, serial data were studied by several methods. The results suggested a trend to increased levels in serum with prolonged therapy, but were not statistically significant. All 168 cerebrospinal fluid (CSF) samples from meningitis patients contained ≤2.9 μg/ml, and only 6 contained >1 μg/ml. There was no apparent relation between dose, time after dose, site of CSF sampling, or concurrent inflammation and CSF ketoconazole concentration. Neither concentration in serum, toxicity, nor outcome correlated with dose, calculated in milligrams per kilogram at the fixed doses (400-mg increments) under study. Likewise, at the various doses, concentration in serum did not correlate with outcome or toxicity, suggesting that individual drug disposition was not an important factor in outcome or toxicity. Toxicity was reversible, and principal side effects were nausea and vomiting (50%), gynecomastia (21%), decreased libido (13%), alopecia (8%), elevated liver function tests (5%), pruritus (5%), and rash (4%). Gastrointestinal and endocrinologic toxicity were dose related and increased at doses >800 mg. The cumulative percent toxicity requiring discontinuation of drug was 6, 17, 23, and 56% at 400-, 800-, 1,200-, and 1,600-mg doses. Doses of >400 mg are thus markedly more toxic, and efficacy data for nonmeningeal disease have not demonstrated that they are more efficacious.