OBJECTIVE To assess which postoperative patients benefit most from MR arthrography of the knee. METHODS One hundred consecutive MR arthrograms obtained in patients who had previous knee surgery underwent retrospective review in consensus by three radiologists after prospective reading by one of the three radiologists. Criterion on MR arthrograms for a retear was abnormal tracking of an intraarticular dilute gadolinium-based contrast material and saline mixture into the substance of a meniscus. Patients were separated into three groups: those with more than 25% meniscal resection, those with less than 25% meniscal resection, and those with meniscal repair. All 100 patients had preoperative MR images to review directly in conjunction with the postoperative MR images. Fifty-seven of these 100 patients underwent second-look arthroscopy. RESULTS Nine patients had MR findings consistent with avascular necrosis. Nineteen patients had marked degenerative arthrosis in the area of previous surgery. Seven patients had chondral defects or injuries. Twenty-nine patients had clear MR evidence of a meniscal retear without any contrast material injected into the joint. In 32 of the 100 patients, intraarticular contrast material was useful in demonstrating a retear. Of these 32 patients, 22 had MR arthrographic evidence of a retear, while 10 had no clear MR arthrographic finding to explain postoperative pain. Four additional patients had no clear MR imaging or MR arthrographic abnormality. All patients with meniscal repair (n = 16) needed MR arthrography to diagnose a residual or recurrent meniscal tear. No patient with less than 25% meniscal resection (n = 23) needed MR arthrography to demonstrate a residual or recurrent meniscal tear. Sixteen of 61 patients with more than 25% meniscal resection needed MR arthrography to demonstrate a residual or recurrent meniscal tear. CONCLUSIONS All patients with meniscal repair required MR arthrography. All patients with meniscal resection of more than 25%, who did not have severe degenerative arthrosis, chondral injuries, or avascular necrosis required MR arthrography. Patients with less than 25% meniscal resection did not need MR arthrography.