The purpose of this study was to determine how the mechanical efficiency of skeletal muscle is affected by level of activation. Experiments were performed in vitro (35 °C) using bundles of fibres from fast-twitch extensor digitorum longus (EDL) and slow-twitch soleus muscles of mice. Measurements were made of the total work and heat produced in response to 10 brief contractions. Mechanical efficiency was the ratio of total work performed to (total heat produced + work performed). Level of activation was varied by altering stimulation frequency between 40 and 160 Hz. Efficiency did not differ significantly between the two muscle types but was significantly lower using 40 Hz stimulation (mean efficiency ± SEM, 0.092 ± 0.012, n = 12, averaged across EDL and soleus) than at any of the other frequencies (160 Hz: 0.147 ± 0.007, n = 12). Measurements of the partitioning of energy output between force-dependent and force-independent components enabled calculation of the amount of Ca²⁺ released and number of cross-bridge cycles performed during the contractions. At 40 Hz stimulation frequency, less Ca²⁺ was released than at higher frequencies and fewer cross-bridge cycles were performed. Furthermore, less work was performed in each cross-bridge cycle. It is concluded that skeletal muscles are less efficient at low levels of activation than when fully activated and this indicates that level of activation affects not only the number of cycling cross-bridges but also the ability of individual cross-bridges to perform work.