BACKGROUND: Direct measurement of blood flow output has been incorporated into ventricular assist devices (VADs), but long-term reliability of the additional device components has raised concerns regarding sensor drift and failure. As an alternative approach, the HeartMate II axial VAD (Thoratec Corp, Pleasanton, CA) estimates device flow output from power consumption and rotational speed of the device motor. This study evaluated the accuracy of HeartMate II flow estimation at the time of implantation. METHODS: In 20 patients, intraoperative blood flow measurement of the HeartMate II flow estimator was compared with flow values obtained with an ultrasonic flow probe placed around the device outflow graft. Estimated and measured VAD flow data were simultaneously recorded and digitally stored while the device motor speed varied from 7,800 to 11,000 rpm and while achieving device flow outputs of 2 to 7 liters/min. Estimated and measured flows were compared using linear regression analyses and root mean square error. RESULTS: HeartMate II flow estimation (FE) demonstrated a linear correlation with ultrasonic flow probe (FP) measurements: FE = 0.74 FP + 0.99 (R(2) = 0.56, p = 0.0001). A root mean square error of 0.8 liters/min was observed between flow estimation and direct flow measurement and suggests a 15% to 20% difference at flows of 4 of 6 liters/min. CONCLUSIONS: These results suggest that HeartMate II flow estimation may be used to provide directional information for trend purposes rather than absolute values of device blood flow output. Patient management should include but not be limited to this information.