Weighted process criteria based on chart review using an algorithmic method, and weighted outcome criteria based on a telephone interview with the patient, were used to assess the quality of emergency room care of women with symptoms of acute urinary tract infection. Outcomes at a municipal hospital were significantly better than those at an affiliated voluntary hospital. Process and outcome scores were positively correlated at both hospitals; the data suggest, but do not themselves adequately test, the hypothesis that the correlation is curvilinear and that there is a threshold process score below which a poor outcome is extremely likely. Our findings suggest that weighted algorithmic assessment criteria are a useful method for quality assessment and are consistent with the testable hypothesis that the greatest improvements in outcome may result from raising the quality of care from a poor to an adequate level rather than from an adequate to an optimal level.