To determine the efficacy and toxicity of two systemically active antifungal agents in the treatment of buccal and oesophageal candidiasis 111 HIV-infected patients with microscopically-confirmed candidiasis were randomized to receive either 200 mg itraconazole once a day or 200 mg ketoconazole twice a day for 28 days in a double blind study. After 1 week of treatment, 75 and 82% of the patients on itraconazole and ketoconazole, respectively, had responded clinically. After 4 weeks of treatment, this had risen to 93% in each group. One patient discontinued itraconozole because of toxicity (rash), five patients discontinued ketaconazole (two nausea, two hepatotoxicity and one rash). Despite successful clinical and mycological clearance, 80% patients had a further episode of candidosis within the next 3 months.