Introduction: There is increasing interest among physiatrists in using bedside ultrasonography to assess rotator cuff tears. Objective: To conduct a method comparison between ultrasonography performed by a single physiatrist at bedside and two validated expert musculoskeletal radiologists at an imaging center. Design: Prospective, blinded comparison study. Setting: Academic outpatient clinic and imaging center. Patients: Seventy-two unilateral shoulders were scanned. Inclusion criteria included pain or weakness with rotator cuff testing and compliance with repeat ultrasonography. Interventions: Ultrasonography performed by the physiatrist was done at bedside during the patient's clinical visit, while the radiologists' scan was performed afterwards in an imaging center. The radiologists trained the physiatrist who was performing the scans. Main Outcome Measurements: The primary outcome was integrity of the rotator cuff (intact, partial tear, full tear). When a posterior cuff (supraspinatus, infraspinatus, teres minor) tear was detected, measurements of length, width, and distance from the biceps tendon were taken. Results: With use of the radiologists' scan as a criterion standard, bedside ultrasonography performed by the physiatrist for detection of posterior cuff tears had a percent perfect agreement of 72.2% for categorization as no tear, partial tear, or full tear. When evaluating dichotomously for presence of a full tear, sensitivity was 82.1% and specificity was 93.9%. Seven (18%) full-thickness tears were missed, and all were essentially small (<15 mm). When physiatrist and radiologist measurements were compared, the mean ± standard deviation (SD) difference in length was 3.4 ± 4.7 mm, width was 2.7 ± 6.7 mm, and distance to the biceps tendon was −0.8 ± 7.5 mm. Conclusions: Office-based bedside ultrasonography is a reasonable modality to rule out medium/large full-thickness posterior cuff tears. However, physicians should be aware that a percentage of small full-thickness tears can be missed. Further imaging should be considered in suspected partial tears and full tears that may be appropriate for surgical repair.