OBJECTIVE: This study aimed to explore the accuracy of three established approaches for legal age threshold estimation in a Croatian sample. MATERIALS AND METHODS: Olze's third molar eruption stages, Haavikko stages, and Demirjian stages were applied in 583 orthopantomograms of Croatian children and adolescents aged 10.00-20.99 years old. The left mandibular second and third molar were assessed. Tooth formation and eruption chronology were established. Receiver operating characteristic (ROC) curves were performed to assess the classification ability of predictive variables for estimating 16- and 18-year-old thresholds. RESULTS: Havikko stage, Demirjian stage and eruption stages in tooth 37 performed poorly for 16-year-old and 18-year-old thresholds. Tooth 38 provide value information for these two age thresholds estimation. For 16-year-old threshold, Haavikko Stage Rc and Demirjian Stage G both yielded Sp over 98% and Bayes PTP over 95% in both sexes. yielded Sp over 98% and Bayes PTP over 95% in both sexes. Eruption stage C yielded Sp over 98%, Bayes PTP over 96% in males, and lower in females (Sp 96.7%, Bayes PTP 88.2%). For 18-year-old threshold, Haavikko stage Ac and Demirjian stage H both yielded 100% Sp and Bayes PTP. Eruption stage D yielded Sp over 96%, Bayes PTP over 90% in both sexes, slightly higehr in females than males. Mean age of tooth 37 at Haavikko stage Ac, Demirjian stage F, H and eruption stage D was statistically lower in females (p<0,05). CONCLUSION: Croatian population-based data on dental maturity were presented. Haavikko stage Rc, Demirjian stage G, eruption stage C and respective advanced stages in tooth 38 are effective for 16-year-old threshold classification. Haavikko stage Ac, Demirjian stage H and eruption stage D are effective for 18-year-old threshold classification.