Biomarkers are being sought that could serve as surrogate end points for chemoprevention trials. Micronuclei, cytoplasmic fragments of DNA, have been proposed as a biomarker and studied in oral pre-malignancy. This study evaluated micronuclei frequency in a randomized chemoprevention trial of oral pre-malignancy. A recent clinical trial evaluated the responses of pre-malignant oral lesions to 3 months of therapy with isotretinoin followed by 9 months of either low-dose isotretinoin or beta-carotene. For 57 study participants, micronuclei were counted in mucosal scrapings of the lesion and in normal-appearing mucosa at baseline and following 3 months and 12 months of therapy. Micronuclei counts were higher in scrapings from the lesion than in the normal-appearing mucosa. Following 3 months of isotretinoin, the micronuclei counts in scrapings of the lesion were significantly reduced. With treatment, the mean micronuclei count declined at 3 months. In a randomized comparison, both isotretinoin and beta-carotene maintained the suppression of micronuclei. The change in micronuclei count was not associated with the clinical or histological response to treatment. Chemoprevention treatment with isotretinoin led to a reduction in frequency of micronuclei, a marker of recent DNA injury, which was then maintained by both isotretinoin and beta-carotene.