Dysplasia grading is widely used to assess risk of transformation in oral potentially malignant disorders despite limited data on predictive value. DNA ploidy analysis has been proposed as an alternative. This study examines the prognostic value for both tests used in a routine diagnostic setting to inform clinical management. A retrospective study of conventional dysplasia grading was conducted on 1,401 patients. DNA ploidy analysis was conducted on a subset of 273 patients and results correlated with clinical information, pathologic diagnosis, and outcome over 5 to 15 years. Malignant transformation occurred in 32 of 273 patients (12%) and, of these, 20 (63%) of preexisting index lesions were aneuploid. Of 241 patients not developing carcinoma, only 39 (16%) of index lesions were aneuploid. Epithelial dysplasia correlated with DNA ploidy status (P < 0.001). The overall positive predictive value for malignant transformation by DNA aneuploidy was 38.5% (sensitivity 65.2% and specificity 75%) and by severe dysplasia grade 39.5% (sensitivity 30% and specificity 98%). DNA diploid and tetraploid status had negative predictive value of 90% to 96%. Combining DNA ploidy analysis with dysplasia grading gives a higher predictive value than either technique alone. Each of three traditional dysplasia grades predicts a significantly different risk of carcinoma development and time to transformation. DNA ploidy analysis had equivalent predictive value and also detected additional risk lesions in the absence of dysplasia.