Estimating chronological age from dental radiographs is a core component of forensic and anthropological investigations. Most existing methods require assessment of multiple teeth within a single jaw or across both jaws. However, such approaches pose a problem when applied to incomplete dentitions, such as in archaeological remains or forensic cases where only one or a few teeth may be preserved. In these contexts, the ability to accurately estimate age from a single tooth becomes highly valuable. This study aimed to identify which tooth, or which combination of teeth, and at which stage of development is best at predicting age for 3 to 16-year-olds.

Our findings indicate that the predictive strength of teeth varies across age categories; different teeth were stronger predictors for different age categories. Specifically, no single tooth consistently emerged as the best predictor throughout the entire age range of 3 to 16 years. This variability may reflect the natural progression of dental development and the biological overlap (collinearity) between stages of adjacent or contralateral teeth. Notably, the second permanent molars demonstrated the highest coefficient of determination (R2) values with smallest confidence intervals suggesting their superior predictive performance for this sample of age range 3 to 16 years [17]. These findings align with previous research that emphasized the value of tooth-specific approaches, such as the work by Liversidge et al. [12], which also found increased accuracy when using maturity scores of individual teeth rather than mean of multiple teeth [12].

The performance of an age estimation method over a wide age range test sample is often expressed as a single value. A recent systematic review and meta-analysis of the accuracy of age estimation using the London Atlas found a standardised mean differences of 0.02 years for MD and 0.78 for MAD [9]. This is of interest,however, our method which evaluates specific tooth stages by age category, allows for more precise age estimation, particularly valuable when working with individual cases rather than population-level trends. The most accurate tooth stages per age category are useful when assessing dental maturity in a living child, but if age is unknown, then it is more useful to assign tooth stages of an individual and refer to which stages are most accurate in Table 4. T his enhances decision-making by linking morphological assessment directly to statistically grounded age estimates.

Our findings also contribute to the ongoing conversation about individual versus group accuracy. While group metrics provide general benchmarks, they are not directly transferable to individual assessments. In this study, several individual tooth stages—including the second molars and maxillary lateral incisor root development—showed low bias and small MAD values for age categories 3 to 12 years, supporting their diagnostic reliability. This may in part be due to the overlap between the age range of our test sample and the critical stages of second molar development.

However, our study is not without limitations. First, the Atlas method—in common with all methods based on age specific illustrations—estimates age as the midpoint of an age category rather than a precise point estimate. Another limitation is the choice of Moorrees’ tooth stages that include crown and root fractions. Tooth stages are discrete events in a continuum of the maturational process and information from longitudinal radiographs about the duration of individual tooth stages is largely undescribed. Some tooth stages are difficult to identify with certainty. Crown or root fractions must be subjectively estimated based on prior knowledge of normal crown height or root length. This can be challenging, particularly when a developing tooth is distorted or not entirely clear radiographically. Although experienced observers may mitigate this variability, it remains a source of potential error. Another limitation of our study is the pooling of males and females, although Seselj et al. [16] have shown that sex does not greatly influence age estimation. In addition, pooled-sex and opposite-sex reference data were almost identical to sex-specific reference data in age estimation performance [11]. Finally, the statistical limitation of evaluating all permutations of tooth combinations adds complexity to interpreting which combinations genuinely offer improved predictive accuracy, particularly in the presence of multicollinearity.

Despite these limitations, our findings underscore a key conclusion: no single tooth predicted age best across all age categories for our test sample. Different teeth were accurate for different age categories. The second permanent molar in both jaws demonstrate strong potential as reliable indicators for chronological age estimation during the mixed dentition period and should be prioritized in subsequent forensic research.