Knowing the age of individuals is crucial for almost any analysis of population dynamics, evolution, palaeontology, management and conservation. The aim of this study was to provide the practitioner with a practical and cost-effective method to estimate the age of large numbers of red deer samples. Using 694 mandibles of Scottish red deer of known age, we compared the bias and precision of five of the most widely used methods for estimating the age of red deer based on tooth characters. Two methods based on reference collections of photographs representing different stages of tooth wear, two methods describing the traits that characterise different classes of the tooth wear and age, and one method based on counting the cementum layers of the radicular pad of the first permanent molar, were used. We also described 13 age classes up to 38 months of age based on different stages of eruption of lower molar teeth. We applied a sequential stepwise-like selection procedure in conjunction with cross-validated predictions using the prediction error sum of squares statistic, with the aim of reducing the large number of traits that the two methods based on tooth wear trait descriptions require to estimate age. We were able to reduce the number of traits by 70% and still gain precision and reduce bias in the predictions, which indicated that the equations provided by these two methods overfitted the age of our reference samples. The cementum layers method was the most precise and least biased of all the methods, followed by Dudley's method. We provide the practitioner with recommendations to allow estimation of the ages of Scottish red deer, together with comprehensive graphic material to facilitate the use of the different methods.