A double-blind, randomized study comparing isotretinoin gel (Isotrex), its vehicle base, and benzoyl peroxide was performed on 77 patients with mild to moderate acne vulgaris. The effect of treatment was assessed by acne grade and lesion count. The vehicle base had no effect, but both active groups produced significant improvements. Benzoyl peroxide and isotretinoin significantly reduced non-inflamed lesions at 4 (P < 0.05), 8 (P < 0.01), 12 (P < 0.01) weeks. Benzoyl peroxide had a more rapid effect on inflamed lesions, their being significant reductions at 4, 8 and 12 weeks (P < 0.01), whereas with isotretinoin there was a significant improvement at 12 weeks (P < 0.01). In addition, compared to placebo, both active treatments significantly reduced inflamed and non-inflamed lesions. Acne grade had improved significantly in the benzoyl peroxide group by 4 weeks (P < 0.01) and in the isotretinoin group by 8 weeks (P < 0.05). No significant change in haematological or biochemical parameters occurred. An irritant dermatitis occurred equally with both treatments but was well tolerated by the patients. This data confirms the clinical benefit of benzoyl peroxide in acne. The initial effect of isotretinoin on non-inflamed lesions in this study suggests that the prime mode of action is on comedone formation or separation whereas benzoyl peroxide has an effect on both comedones and inflammation.