Modelling the mechanical effect of muscles is important in several research and clinical contexts. However, few studies have investigated the effect of different muscle discretizations from a mechanical standpoint. The present study evaluated the errors of a reduced discretization of the lower-limb muscles in reproducing the muscle loading transferred to bones. Skeletal geometries and a muscle data collection were derived from clinical images and dissection studies of two cadaver specimens. The guidelines of a general method previously proposed for a different anatomical district were followed. The data collection was used to calculate the mechanical effect of muscles, i.e. the generalized force vectors, and the errors between a large and a reduced discretization, in a reference skeletal pose and in the extreme poses of the range of motion of joints. The results showed that the errors committed using a reduced representation of muscles could be significant and higher than those reported for a different anatomical region. In particular, the calculated errors were found to be dependent on the individual anatomy and on the skeletal pose. Since different biomechanical applications may require different discretization levels, care is suggested in identifying the number of muscle lines of action to be used in musculoskeletal models.