In long-term studies of wild animals, individuals are often caught initially as adults, and so their age is unknown. To better understand age structure, cohort effects and life-history traits, it is desirable to ascribe approximate ages to individuals. Tooth wear has been used as a proxy for age in many mammals, including the Eurasian badger Meles meles. We used tooth-wear data derived from serial captures of over 2000 badgers of known age to calibrate the relationship between tooth wear and age and produce a predictive model. As badgers were recaptured throughout their lifetime, we used all observations of tooth wear from each individual of unknown age to estimate its year of birth. By taking into account repeated observations of tooth wear, we generated more accurate and internally consistent predictions. Spatial variations in the rates of tooth wear are likely to relate to differences in the diet and more rapid rates of wear among male badgers may be linked to higher levels of food intake. The performance of the optimum model at accurately predicting badger age from tooth wear was assessed using data from known-age animals. The reliability of predictions declined with age but for our study population, there was an 88% probability of being accurate to within 1 year. The model performed less well at predicting age from a single observation (71% accuracy to within 1 year) than from repeated observations of tooth wear. Individuals of unknown age are likely to be encountered in most studies of free-living animal populations, and in many cases, there will be physiological indicators (such as tooth wear in mammals) that can be used to approximate age. Combining repeated observations of these indicators in unknown-age individuals is likely to improve the accuracy of estimates of population parameters.