The study was designed to investigate the accuracy and precision of three different methods for estimation of chronological age (Liliequist and Lundberg, 1971; Demirjian ci' al., 1973 and Gustafson and Koch, 1974), based on tooth formation. Accuracy test. For each method, the mean difference and degree of linear correlation between estimated and true chronological age was determined in 300 3.5-12.5-year-old Swedish children. The data were arranged in subgroups according to sex and age. Precision test. For each method, the inter-examiner variation was studied in terms of the mean difference between the estimated age obtained from independent readings of 60 panoramic radiographs of the teeth by the two examiners. The intra-examiner variation was studied by calculating the mean difference between two independent readings of the same 60 radiographs by the same examiner. A high accuracy was found when the method proposed by Demirjian et al., was applied to 3.5-6.5-year-old children. This method showed a low accuracy in the older age-groups, however. The accuracy of the method devised by Liliequist and Lundberg (1971) was found to be low in all age-groups, and age determination using this method resulted in systematic underestimation of age. The accuracy of the method described by Gustafson and Koch (1974) was high when applied to boys, but low in girls. The precision was found to be high for the methods of Demirjian et al. (1973) and Liliequist and Lundberg (1971) but somewhat lower for the method of Gustafson and Koch (1974). Estimation of age is preferably done during early childhood. Of the methods tested, the one proposed by Demirjian et al. (1973) is the most reliable at these ages, due to its comparatively high accuracy and precision.