Background: : Although several studies have described magnetic resonance imaging findings in shoulders of asymptomatic volunteers, no such investigation has been performed on asymptomatic dominant and nondominant shoulders of elite overhead athletes. HYPOTHESIS: Asymptomatic dominant shoulders of elite overhead athletes may have a higher incidence of magnetic resonance imaging abnormalities than either their nondominant shoulder or shoulders of asymptomatic volunteers. STUDY DESIGN: Prospective cohort study. METHODS: Detailed magnetic resonance imaging scans of asymptomatic dominant and nondominant shoulders of elite overhead athletes were obtained. Three experienced musculoskeletal radiologists interpreted each scan for multiple variables, including rotator cuff appearance. Images from a surgical control group were intermixed to assess accuracy and control for observer bias. A 5-year follow-up interview was performed to determine whether magnetic resonance imaging abnormalities found in the initial stage of the study represented truly clinical false-positive findings or symptomatic shoulders in evolution. RESULTS: Eight of 20 (40%) dominant shoulders had findings consistent with partial- or full-thickness tears of the rotator cuff as compared with none (0%) of the nondominant shoulders. Five of 20 (25%) dominant shoulders had magnetic resonance imaging evidence of Bennett's lesions compared with none (0%) of the nondominant shoulders. None of the athletes interviewed 5 years later had any subjective symptoms or had required any evaluation or treatment for shoulder-related problems during the study period. CONCLUSIONS: Magnetic resonance imaging alone should not be used as a basis for operative intervention in this patient population.