To examine how patient experiences of the interpersonal aspects of quality of care varied among a group of 15 New York City hospitals, and the extent to which hospital and patient characteristics explained interhospital variability, a telephone survey was conducted with 3,423 randomly selected patients discharged from 15 New York City hospitals. Bivariate analysis, multiple linear regression, and least square means were used to assess the effects of 5 hospital characteristics and 15 patient characteristics on reports about problems with care. Outcome measures included patients' reported problems with selected aspects of care, patients' ratings of care, and patients' willingness to recommend the hospitals from which they had been discharged. The 15 hospitals varied widely in the rates at which patients reported problems with their care (10.7-21.7, mean = 14.8, p < 0.001). A multivariate model showed that patients in fair or poor health, those without a regular doctor, younger patients, and minorities other than black and Hispanic were more likely to report problems with aspects of their care. Medicaid volume was also a strong, significant predictor of problem scores. Patient reports can be used to measure differences in quality of interpersonal care among hospitals. Only some of these differences are explained by patient and hospital characteristics, indicating that other factors facilitate or inhibit the delivery of high-quality interpersonal care.