Accurate preoperative prediction of functional donor and remnant hemiliver volumes in live donor liver transplantation is essential in preventing postoperative liver failure and optimizing safety. Our aim was (1) to evaluate volume variability associated with multiphasic CT imaging and (2) to determine over- or under-estimations of 3-D CT graft-volume assessments based on 'largest' versus 'smallest' CT phases with respect to intraoperative findings. Native, arterial and venous phase CT images from 83 potential live liver donors were subject to 3-D CT liver volume calculations and virtual 3-D liver partitioning. Estimates were compared to intraoperative volumes obtained in 43 cases. Calculated (preoperative) graft-volume-body-weight-ratios (GVBWR) versus measured (intraoperative) graft-weight-body-weight-ratios (GWBWR) were analyzed. Significant differences in total liver volume- and in graft-liver volume calculations were found among the largest (venous) and smallest (native) CT phases. High significant overestimations were observed in graft-volume determinations and in GVBWR-calculations based on the 'largest' CT phase when compared to intraoperatively obtained graft-weight and -GWBWR values. In contrast, differences among intraoperative measurements and preoperative calculations based on the 'smallest' CT phase yielded less significant overestimations. While 3-D CT volumetry based on the 'largest' (venous) CT phase is associated with considerable overestimation, 3-D volumetry based on the 'smallest' (native) CT phase accurately matches the intraoperative findings.