The efficiency of positive work was measured for rat medial gastrocnemius muscle at 25 degrees C during repeated contractions. Six muscles were stimulated to perform concentric contractions preceded by an active prestretch (PS contractions) and six muscles made to give concentric contractions from an isometric state (PI contractions). Both lengthening and shortening of the muscles (distance: 6 mm) occurred at a constant velocity of 20 mm s-1 (1.5 fibre lengths s-1). Stimulation was started 150 ms prior to the onset of concentric contraction for both types of contraction. For the PS contractions this meant that the active state was developed during the last 2.4 mm of the lengthening. Energy consumption (calculated from high-energy phosphate consumption) appeared to be equal for both types of contraction, although positive work output was 39.4% higher in the PS contractions than in the PI contractions. The efficiency of positive work was 36.8 +/- 3.5% in the PS contractions and 26 +/- 2.0% in the PI contractions. In contrast to results of previous studies, the positive work done by the muscle in the PS contractions was much larger than the negative work done on the muscle during stretch owing to the applied stimulation protocol which was intended to simulate in vivo conditions during running. The efficiency of positive work in the PS contractions is too low to explain the efficiencies of 40-70% reported for human and animal running.