The objective of the study was the determination of the intra- and interobserver agreement of three magnetic resonance imaging (MRI) criteria for reparability: a peripheral rim smaller than 4 mm, a tear longer than 10 mm, and homogenous aspect of meniscal tissue. In two rounds with an interval of at least 6 weeks, three orthopedic surgeons and three musculoskeletal radiologists studied the preoperative MRI scans of 63 patients with a longitudinal full-thickness medial or lateral meniscal tear. All patients had an arthroscopic meniscal repair. The blinded images were evaluated measuring the tear length and rim width and meniscal aspect was classified. Agreement was calculated using the linear-weighted kappa coefficient (κ) and the intraclass correlation coefficient (ICC). Examiner agreement strength was defined according to the guidelines of Landis and Koch. Intraobserver agreement was poor to good (κ, 0.12-0.72) for the classification of the meniscal aspect and decreased in lateral meniscal tears. The interobserver agreement for meniscal aspect was mainly poor to fair (κ, 0.09-0.53). The intraobserver reliability for measurement of the length of the meniscal tear was moderate to excellent (ICC, 0.51-0.80) for all observers in both rounds and moderate to good (ICC, 0.59-0.73) for measurement of the peripheral rim width. The interobserver agreement on tear length and rim width was moderate in both rounds (ICC, 0.58 and 0.50 in round 1; 0.50 and 0.50 in round 2, respectively). Tear length and rim width are the only two measurements with moderate to good agreement. However, these measurements do not predict reparability of longitudinal meniscal tears on MRI images.