Observational studies have demonstrated a positive association between phenytoin use and HDL cholesterol (HDL-C). Our goal was to determine whether phenytoin raises HDL-C in nonepileptic subjects at risk for coronary artery disease. We performed a double-blind, placebo-controlled, parallel-group study in 41 subjects with reduced levels of HDL-C. Subjects were placed on an American Heart Association Step I diet and were randomized to receive either phenytoin or placebo for 3 months. Serum levels of phenytoin were monitored and adjusted to between 7.5 and 15 micrograms/mL. Fasting levels of lipids and lipoproteins were determined twice at baseline (weeks -2 and -1) and during the treatment phase of the study (weeks 11 and 12). Compared with dietary baseline, phenytoin-treated subjects experienced significant paired percent increases in total HDL-C (12.4%; P < .01), an effect confined to the HDL2 subfraction (137%; P < .01). The paired percent increases in HDL-C and HDL2 levels remained significant after adjustment for placebo (P < .05, P < .025, respectively). There were no significant differences in the paired percent changes from dietary baseline in total cholesterol, triglyceride, or LDL cholesterol levels between placebo and phenytoin-treated groups. The significant paired percent increases in total HDL-C and HDL2 from dietary baseline suggest a potential role for phenytoin in subjects with reduced levels of HDL-C.