This study was performed to compare time-of-flight MR angiography in detecting varices with conventional portography and endoscopy in patients with chronic liver disease and to compare MR tomographic images with projection angiograms. In eight patients findings on conventional arterial portography (considered the gold standard) were compared with the MR findings. Varices were graded by size and extent on a scale of 0-3. Splenic varices were detected in all patients by MR but were scored larger on portography in 6 (55%) of 11 comparisons. All left gastric varices were identified at MR and were rated within one grade of those seen at portography. All esophageal varices were identified at MR. Among 12 patients with upper endoscopy, MR rated varices significantly (p less than 0.05) larger than endoscopy in 8 (67%) of 12 comparisons. Varices were detected in two patients on MR that were not seen endoscopically. Extraperitoneal varices were identified in six (75%) of eight patients on MR and were only visualized in one patient at portography. Varices were equally well detected using either axial or coronal images. No significant difference existed when tomographic images were compared with projection images. Time-of-flight MR angiography is a valid technique for noninvasive imaging of abdominal varices. Tomographic single slice images are generally as good as projection images except possibly in the demonstration of extraperitoneal varices.