This paper explores the relationship between cost and quality of hospital care. A total operating cost function is estimated for 137 US Department of Veterans Affairs hospitals for 1988–1993 using three rate-based measures of quality as regressors. The high likelihood of the existence of measurement error in quality in the cross section leads to the application of novel instrumental variable techniques. Results suggest that mortality and readmission indices are adjusted inadequately for illness severity. The measure on the failure to follow up inpatient discharges with outpatient care, however, appears to increase cost. The results of this paper underscore a number of practical difficulties and challenges facing government or other systems in evaluating the relative performance of their hospitals. Copyright © 1999 John Wiley & Sons, Ltd.