Left heart bypass (LHB) was performed in goats with normal ventricles and in those with LV failure. A pneumatic type pusher-plate pump can be operated in fixed rate, synchronous mode variable pulse rate (VR). In this experiment, the pump was operated in VR, that is, the pusher-plete position signal from a hall effect sensor was utilized to regulate the pump stroke at a constant level, but its rate was allowed to vary automatically depending on the preload and afterload. LHB was performed in 12 goats with normal ventricles. Blood was drained from the left atrium and was returned to the descending aorta. Pump flows averaged 61.3% of cardiac output (CO). When bypass was started, CO decreased from 72.5 to 57.2 ml/min/kg (p<0.005), and LV(max dp/dt) decreased from 3943 to 2989 mmHg/sec (p<0.05). Atrial pressure (AP), A-VDO(2), pulmonary atrial pressure (PAP) and TTI did not show significant changes. Regional myocardial blood flow (RMBF), which was measured using the hydogen gas clearance method, decreased from 128 to 85 ml/min/100g (p<0.05). In 12 goats, LV failure was induced by injecting 0.08 ml/kg of 5N NaOH into the LV free wall. The weight ratio of myocardial necrosis to the LV free wall was 22.7%. Pump flows averaged 52.5% of CO. When bypass was started, CO increased from 45.7 to 61.4 (p<0.005). AP increased from 85.2 to 124 (p<0.001), A-VDO(2) decreased from 6.08 to 4.74 vol% (p<0.01) and PAP decreased from 33.2 to 30.3 (p<0.02). LV(max dp/dt) and TTI did not show significant changes. RMBF, which was determined at normal LV free wall, decreased from 98.6 to 76.1 (p<0.05). These data suggest that LHB is more effective in goats with LV failure than in those with normal ventricles in terms of left ventricular and systemic circulatory support.