Prediction model of one repetition maximum (1RM) based on anthropometrical characteristics on male and female. Brazilian Journal of Biomotricity. v. 7, n. 1, p. 43-52, 2013. The goal of the present study was to determine a model for predicting one maximum repetition (1RM) based on anthropometrical characteristics for exercises leg abduction (LA) and leg curl (LC) on males and females. Forty-eight volunteers trained in strength training were submitted to an anthropometric evaluation, followed by a 1RM test for exercises LA and LC. After resting for at least 48h, they randomly underwent either a new 1RM. The model’s for predicting 1RM strength were obtained by multiple linear regression for LA and LC exercises assuming as independent variables: fat-free mass and gender. The best model’s explained 66% and 78% of the known variance and have a standard error of estimate of 15.5% and 15.4% of the measured 1RM for LA and LC, respectively. The measured 1RM were 54.4 ± 14.3 Kg and 54.0 ± 17.4 Kg and predicted 1RM were 54.4 ± 11.8 Kg and 54.0 ± 15.4 Kg, with the difference between achieved and predicted 1RM in mean 0.0 ± 8.3 Kg and 0.0 ± 8.1 Kg for LA and LC, respectively (P < 0.01). The obtained model’s showed an acceptable validity, and further can be currently used as a tool for predicting the 1RM workload.