Abstract： Background Multislice helical computed tomography (MSCT) has been used to depict coronary anatomy noninvasively, and proved useful for evaluating ventricular function. The aim of our study was to assess the accuracy of ventricular volume as measured by MSCT.Methods Fourteen human left ventricular (LV) and 15 right ventricular (RV) casts were scanned by MSCT. A series of LV and RV short-axis images were reconstructed later with slice thickness of 2.0 mm, 3.5 mm, 5.0 mm, 7.0 mm, and 10.0 mm. Ventricular volume was calculated by the multislice tomographic Simpson's method. True LV and RV cast volumes were determined by water displacement. Results Both calculated LV and RV volumes correlated highly with the corresponding true volumes (all r>0.95, P<0.01). But with slice thickness from 2.0 mm to 10.0 mm, MSCT scanning overestimated the corresponding true volume by (3.21±5.95) ml to (12.58±8.56) ml for LV and (10.22±8.45) ml to (23.91±12.24) ml for RV (all P<0.01). There was a very high correlation between the overestimation and the selected slice thickness for both LV and RV volume measurements (r=0.998 and 0.996, P<0.01, respectively). However, when slice thickness was reduced to 5.0 mm, the overestimation for both LV and RV volume measurements became nonsignificant for slice thickness from 2.0 mm to 5.0 mm. Conclusions Both LV and RV volumes can be accurately estimated by MSCT. Thinner slice has more accurate calculated volume. However, 5.0 mm slice thickness is thin enough for an accurate measurement of LV or RV volume.