The aim of this study was to test the performance of the third molar maturity index (I3M) for discriminating between adults and minors. Panoramic radiographs of 760 individuals aged 11 to 25 years were assessed and validity of I3M < 0.08 was tested. A receiver operating characteristic (ROC) curve analysis and Youden's index were used to test the performance of the method. The sensitivity and specificity were 0.554 (95% confidence intervals [CI], 0.487-0.619) and 0.962 (95% CI, 0.926-0.983) in females, 0.750 (95% CI, 0.666-0.829) and 0.949 (95% CI, 0.910-0.974) in males, respectively. The proportion of correctly classified individuals was 0.747 (95% CI, 0.703-0.786) and 0.884 (95% CI, 0.815-0.895) in females and males, respectively. In addition, an optimal cut-off value was determined as I3M < 0.19 for females. When we used this cut-off value in females, sensitivity, specificity and accuracy were as follows: 0.820 (95% CI, 0.764-0.867), 0.833 (95% CI, 0.775-0.880) and 0.826 (95% 0.768-0.843). The I3M gave better results when using a sex-specific cut-off value in a Turkish sample.