OBJECTIVE Nontumorous perfusion defects occur during CT arterial portography (CTAP) as normal variants or in cirrhosis, focal fatty infiltration, and portal vein obstruction. The purpose of this study was to determine whether delayed CT 4-6 hr after CTAP improves sensitivity to hepatic tumors and differentiates them from other hepatic perfusion defects. METHODS CTAP was done at 1-cm increments on 26 patients for evaluation of hepatic tumors. Delayed CT scans were obtained 4-6 hr later in all patients. Two observers retrospectively reviewed the CT scans obtained during CTAP and recorded size, shape, and location of suspected hepatic tumors. Confidence levels were assigned for each tumor. The delayed CT scan was then interpreted in conjunction with the CT scans obtained during CTAP, and confidence levels were reassigned. Surgical correlation was obtained for all patients. In the 26 patients, 86 masses were found at surgery. The sensitivity and number of false-positives for both CTAP alone and CTAP combined with delayed CT were compared with a two-tailed Student t-test. Receiver-operating-characteristic analysis also was performed. RESULTS CTAP detected 73 (85%) of the 86 hepatic masses. Delayed CT had no effect on the sensitivity of CTAP. However, adding delayed CT decreased the total number of false-positives by 11, a statistically significant difference (p < .05). Receiver-operating-characteristic analysis revealed a significantly greater (p < .05) area under the curve (Az index) of 0.927 +/- 0.025 for CTAP combined with delayed CT compared with 0.886 +/- 0.032 for CTAP alone. Delayed CT was most useful for larger (> 1 cm) wedge-shaped perfusion defects and least useful for smaller (< 1 cm) round defects. CONCLUSIONS Delayed CT has no effect in detecting tumors but may be useful for differentiating tumors from other hepatic perfusion defects seen on CTAP. The greatest benefit of delayed CT is in evaluating regions obscured by large wedge-shaped perfusion defects on CT scans obtained during CTAP.