To determine whether daily use of nystatin pastilles can prevent initial outbreak or recurrence of oral candidiasis in HIV-infected patients and to identify factors associated with outbreaks during 20-week follow-up, a randomized, double-blind, placebo-controlled clinical trial was conducted. Subjects were 128 HIV-infected men (aged 27-60 years) who either had had no documented episode of oral candidiasis in the previous year or had been clinically clear of oral candidiasis for at least 72 h before randomization. Study arms were two placebo pastilles, one nystatin (200,000 U) and one placebo pastille, or two nystatin pastilles daily for 20 weeks. The main outcome measure was time to oral candidiasis, as determined by potassium hydroxide (KOH) smear and fungal culture. A multivariate proportional hazards model showed that four factors were significant (p < 0.001) in predicting time to oral candidiasis: nystatin treatment (hazard ratio 0.59), history of oral candidiasis (3.58), Candida albicans carriage (2.79), and CD4 count at randomization (0.65). In this small group of subjects, nystatin appeared to be effective in delaying onset of oral candidiasis. Patients with CD4 counts < 200 who are carriers of C. albicans and have a history of oral candidiasis may be most likely to benefit from antifungal prophylaxis.