BackgroundPubertal timing is linked to the emergence and severity of psychopathology during adolescence. However, variability in methods used to calculate pubertal timing may contribute to inconsistencies in the literature. The aim of this study was to develop a new measure of pubertal timing and investigate its association with psychopathology.MethodsWe analyzed data from the Adolescent Brain Cognitive Development (ABCD) cohort, a longitudinal study of ∼11,000 children. We implemented three different models of pubertal timing by predicting chronological age from i) observed physical development, ii) hormonal assays, and iii) a combination of physical development and hormones, using a supervised machine learning method. Our method of quantifying pubertal timing was calculated as the difference between predicted and actual age in each model. The performance of the new physical pubertal timing model was also compared to a measure of pubertal timing commonly used in the literature. We compared the three new measures by evaluating their associations with psychopathology.ResultsThe combined model provided the best prediction of age. The physical pubertal timing model had better performance compared to the commonly used existing model. Pubertal timing estimates from both physical and combined models were significantly associated with most dimensions of psychopathology in males and females (early timing associated with higher symptoms). The physical model accounted for more variance in psychopathology than the combined and hormonal models.ConclusionsThis study proposed new models of pubertal timing that utilize multiple pubertal features and account for nonlinear associations with age. Findings suggest that timing of physical maturation may play a predominant role in predicting psychopathology in early adolescence. Further investigation is needed to see if measures of timing that incorporate hormones are more predictive of psychopathology across different stages of late childhood and adolescence.