Objective To evaluate the accuracy of a three-dimensional (3D) magnetic position sensor system in the quantification of ventricular stroke volumes in a dynamic model. Methods A latex balloon model of the left ventricle was suspended in a water bath connected to a pump producing 10 different pulsatile stroke volumes (15-65mi/beat). Scanning was performed using a 5.0 mHz transducer mounted with a Flock of Birds (FOB) magnetic receiver (GE System Five). The probe was scanned to sweep continuously across and over the balloon volume over 3 - 7 seconds. Digital loops were stored on magneto-optical disks and reviewed retrospectively using 3D Echopac software (GE) based on Simpson's method and compared with a two-dimensional (2D) biplane area-length method (1/2 L x R) measurements at end systole and end diastole. Both 3D and 2D derived stroke volumes were compared with the reference stroke volume calculated by direct measurement of balloon capacity. Results There was an improved correlation between 3D stroke volume and reference stroke volume ( y = 0.91 x + 0.41, r = 0.97, SEE = 2.83 ml, P = 0.0001 ) compared to 2D stroke volume and reference stroke volume (y=0.49x+8.68, r=0.87, SEE=3.87 ml, P=0.0011, difference between 2D and 3D P＜0.003). Conclusion 3D magnetic FOB scanning is practical, accurate and should facilitate assessment of left ventricular function.