Eccentric contractions are thought to require a unique neural activation strategy. However, due to greater intrinsic force generating capacity of muscle fibres during eccentric contraction, the understanding of neural modulation of different contraction types during submaximal contractions may be impeded by the force normalisation procedure employed. In the present experiment, subjects performed maximal isometric dorsiflexion at shorter (80°), intermediate (90°) and longer (100°) muscle lengths, and maximal concentric and eccentric contractions. Thereafter, submaximal concentric and eccentric contractions were performed normalised to either isometric maximum at 90° (ISO), contraction type specific maximum (CTS) or muscle length specific maximum (MLS). When using ISO or MLS for normalisation, mean submaximal eccentric torque levels were significantly lower when compared to CTS normalisation (11 and 7% lower compared to CTS; p = 0.003 and p = 0.018 for ISO and MLS, respectively). These experimentally observed differences closely matched those expected from the predictive model. During submaximal concentric contraction, mean torque levels were similar between ISO and CTS normalisation with similar discrepancies noted in EMG activity. These findings suggest that normalising to ISO and MLS might not be accurate for assessment and prescription of submaximal eccentric contractions.