The aim of this study was to examine the significance of evaluating dmft for predicting the incidence of permanent teeth caries in Japanese girls. The subjects were two birth cohorts in a Primary girls' school: cohort I composed of 45 girls born in 1981 and 1982 and cohort II composed of 53 girls born in 1989 and 1990. In both cohorts, there was a significant correlation between the dmft score in the first grade and the DMFT score in the sixth grade. The validity of employing dmft score was examined by receiver operating characteristic (ROC) curve and risk ratios. The shape of the curve and areas under the curve were similar in the two cohorts. In cohort II, sensitivity and specificity for the optimal cut-off level (dmft ≥ 5) were 0.519 and 0.923, respectively. The positive and negative predictive values for that level were 0.875 and 0.649, respectively. In cohort I, sensitivity and specificity were 0.741 and 0.722 for that level and positive and negative predictive values were 0.800 and 0.650, respectively. In both cohorts, the risk ratio for permanent teeth caries was significant for many cut-off levels of dmft. The results indicated that the dmft score is a useful predictor of the permanent teeth caries in Japanese primary school girls. (J. Oral Sci. 48, 233-237, 2006)