OBJECTIVE To assess the relations between clinically recognized lumbar spinal stenosis and the conclusions of masked radiologists and electrodiagnosticians. DESIGN Prospective, masked, double-controlled trial. SETTING University spine center. PARTICIPANTS One hundred fifty persons age 55 to 80 years with or without back pain and with or without magnetic resonance imaging (MRI)-demonstrated stenosis, screened for neuropathy risk, previous surgery, or cancer. INTERVENTIONS Questionnaires on pain and function; ambulation testing and physical examination; and masked electrodiagnotics and MRI. MAIN OUTCOME MEASURE Diagnostic impressions of the examining clinician, radiologist, and electrodiagnostician. RESULTS Following application of post hoc exclusion criteria and elimination of patients due to incomplete or inadequate test data, the clinical diagnosis was lumbar stenosis in 50 subjects, back pain in 44 subjects, and no pain in 32 subjects. Radiologic and clinical impression had no relation (P = .80 vs asymptomatic, P = .99 vs back pain controls). Electrodiagnostic impression trended to relate to clinical impression (P = .14 vs asymptomatic, P = .09 vs back pain). Retrospective application of age-related electrodiagnostic norms for paraspinal electromyographic and limb motor unit changes, established in this study, reclassified 13 of the 17 asymptomatic persons whom the electrodiagnostician thought had stenosis. The clinical impression did correspond to history and physical examination findings typically associated with spinal stenosis and to the independent impression of a neurosurgeon who examined MRI and clinical, but not to the electrodiagnostic data. CONCLUSIONS The impression obtained from an MRI scan does not determine whether lumbar stenosis is a cause of pain. Electrodiagnostic consultation may be useful, especially if age-related norms obtained in this study are applied.