Patient ratings of physician's patient-centered communication are used by various specialty credentialing organizations and managed care organizations as a measure of physician communication skills. We wanted to compare ratings by real patients with ratings by standardized patients of physician communication. We assessed physician communication using a modified version of the Health Care Climate Questionnaire (HCCQ) among a sample of 100 community physicians. The HCCQ measures physician autonomy support, a key dimension in patient-centered communication. For each physician, the questionnaire was completed by roughly 49 real patients and 2 unannounced standardized patients. Standardized patients portrayed 2 roles: gastroesophageal disorder reflux symptoms and poorly characterized chest pain with multiple unexplained symptoms. We compared the distribution, reliability, and physician rank derived from using real and standardized patients after adjusting for patient, physician, and standardized patient effects. There were real and standardized patient ratings for 96 of the 100 physicians. Compared with standardized patient scores, real-patient-derived HCCQ scores were higher (mean 22.0 vs 17.2), standard deviations were lower (3.1 vs 4.9), and ranges were similar (both 5-25). Calculated real patient reliability, given 49 ratings per physician, was 0.78 (95% confidence interval [CI], 0.71-0.84) compared with the standardized patient reliability of 0.57 (95% CI, 0.39-0.73), given 2 ratings per physician. Spearman rank correlation between mean real patient and standardized patient scores was positive but small to moderate in magnitude, 0.28. Real patient and standardized patient ratings of physician communication style differ substantially and appear to provide different information about physicians' communication style.