Currently, there is no consensus on the appropriate treatment for low-grade oral dysplasia. This is mainly due to the difficulty in predicting outcome for this heterogeneous group of lesions. In this study, we constructed a detailed clinical history of 66 mild and moderate dysplasias in order to determine how treatment affected outcome, and to evaluate the effect of treatment on lesions with different genetic profiles, which are defined by patterns of loss of heterozygosity (LOH) associated with low, intermediate and high risk of progression [Clin. Cancer Res., 6, 357-62, 2000]. The results showed that although treatment guided by clinical removal of leukoplakia reduced cancer progression risk in all three risk groups, the amount of reduction in our study group did not reach statistical significance. To assess whether completeness of lesion removal was a major factor in recurrence, repeat biopsies at the primary sites were analyzed for persistent LOH status on chromosomes 3p, 4q, 8p, 9p, 11q, 13q and 17p. Strikingly, eight of 17 cases judged clinically removed contained the same molecular clones in the initial and subsequent biopsies, suggesting incomplete removal. When molecular information was included in the assessment of lesion removal, treatment significantly reduced the risk of progression for cases with intermediate (P=0.043) and high risk (P=0.001) genetic profiles, but not cases with low-risk profiles. A 9.1-fold decrease in progression risk was observed for those with high-risk profile. Altogether, these data suggest the use of molecular profiles to guide the treatment of low-grade dysplasia. Our data also suggest that currently an inadequate margin may in part be responsible for the high rate of recurrence, especially in high-risk lesions.