With magnetic resonance imaging (MRI) the surgeon has such an effective diagnostic tool in the diagnosis of a meniscal lesion that the times of diagnostic arthroscopy should be in the past. A total of 823 patients with clinically diagnosed meniscal lesions were divided into two groups: group A, 143 patients underwent MRI and 75 of those arthroscopy; group B, 680 patients, 201 (30%) of whom were operated after being only clinically examined. MRI was done on a Siemens Medical System Magnetom 1.5 Tesla with a 256 x 256 matrix. Spinal echo and gradient echo images were performed with slices of 2 to 4 mm thickness. All meniscal tears were graded according to Reicher and Crues, respectively. Grades III and IV were judged to be positive for a meniscal lesion. Arthroscopy was carried out under general anesthesia and the usual technique. The MRI revealed the following results: Medial meniscus: accuracy 95%, positive predictive value (PPV) 92%, negative predictive value (NPV) 95%, sensitivity 98%, and specificity 82%, lateral meniscus: accuracy 97%, PPV 92%, NPV 98%, sensitivity 94%, and specificity 98%. The overall values for MRI of the medial and lateral menisci combined were: accuracy 96%, PPV 93%, NPV 98%, sensitivity 96%, and specificity 90%. The clinical examination often failed to diagnose a meniscal lesion: accuracy 64%, PPV 59%, NPV 89%, sensitivity 96%, and specificity 33% for the medial meniscus. For the lateral meniscus the accuracy was 91%, PPV 61%, NPV 98%, sensitivity 89%, and specificity 91%. The overall values for the clinical investigation of the medial and lateral menisci combined were: accuracy 78%, PPV 60%, NPV 94%, sensitivity 93%, and specificity 62%. Investigation of all 201 patients operated from group B with MRI would have cost $160,800. The cost of 30% fewer arthroscopies would have been $562,800--in total $723,600. The operation of all 201 patients cost $804,000. Hence about $80,000 could have been saved by scanning all 201 patients and therefore reduce the rate of diagnostic arthroscopies.