This study defines the accuracy of gadopentetate-dimeglumine-enhanced MR imaging in patients with failed back surgery syndrome by comparing the imaging studies with surgical findings in a large patient population. From June 1988 to March 1989, 193 postoperative patients had MR imaging of the lumbar spine both with and without contrast administration. Of this group, 27 had repeat surgery at 31 levels: these patients comprise the study group. Postcontrast MR diagnoses were as follows: scar only (n = 4), disk only (n = 13), scar and disk (n = 9), and no aberrant epidural tissue (n = 5). Surgical diagnoses differed from the MR diagnoses in two patients at two levels. In one patient, disk was diagnosed on MR while osteophyte was present at surgery. In the other patient, scar only was diagnosed by MR but disk and scar were present at surgery. These data, when combined with the authors' previous experience, give pre- and postcontrast MR a 96% accuracy in differentiating scar from disk in 44 patients at 50 reoperated levels. For patients 6 or more weeks past surgery, sagittal and axial T1-weighted MR imaging before and after administration of gadopentetate dimeglumine is an effective method of evaluating the postoperative lumbar spine.