To determine whether the University of California, Los Angeles (UCLA) shoulder score completed by the patient is comparable to that obtained by the clinician, 110 patients who had undergone either arthroscopic subacromial decompression (ASD) or rotator cuff repair (RCR) completed the UCLA score questionnaire by mail at 2 months (ASD subgroup) or 4 months (RCR subgroup). Patients were then assessed by 2 clinicians in random order. There was a very good level of agreement for the overall UCLA score for the total cohort (N = 100 with complete data) (intraclass correlation coefficient [ICC], 0.910; 95% confidence interval [CI], 0.87 to 0.94) and for ASD (n = 46) (ICC, 0.951; 95% CI, 0.92 to 0.97) and good agreement for RCR (n = 54) (ICC, 0.734; 95% CI, 0.61 to 0.83). Agreement between patient and assessor 1, patient and assessor 2, and assessors 1 and 2 was also very good (with whole-cohort ICCs of 0.875, 0.910, and 0.935, respectively). Bland-Altman plots showed little systematic disagreement and consistently narrow limits of agreement. Patient self-administration of the UCLA shoulder score yields acceptable results.