This pilot study was developed to compare the relative diagnostic accuracies of physical findings, magnetic resonance imaging (MRI) and arthroscopy for internal derangements in knee osteoarthritis (OA) patients. Nine patients with locking and/or giving way in 10 knees underwent MRI and arthroscopy; the relative diagnostic accuracies for meniscal tears were studied and compared with physical findings. Eleven meniscal and no cruciate ligament tears were noted by MRI and/ or arthroscopy. Using arthroscopy as the comparison standard, the sensitivity of MRI for meniscal tears was 33.3%, specificity was 96.6%, and diagnostic accuracy was 75.0%. No significant correlations between physical findings and MRI or arthroscopy findings were found. It seemed that participants with normal radiographs had false positive MRIs more frequently. These preliminary data suggest that physical findings may not be adequate for the diagnosis of meniscal tears in patients with associated knee OA. A larger study may be warranted to further test this hypothesis. Because the presence of a meniscal tear may change therapy toward specific physical therapy modalities and/or meniscal repair, knee OA patients with mechanical symptoms may require an MRI or arthroscopy to establish the presence of a meniscal tear. Further testing is required to confirm the suggestion from these cases that patients with normal or minimally abnormal radiographs may require a diagnostic arthroscopy rather than an MRI to demonstrate a meniscal tear.