In this article, we investigate the relationship between hospital length-of-stay (LOS) and quality of care. We use hospital claims records from Medicare beneficiaries in Michigan to estimate condition-specific models for predicting patients' LOSs. With these models and a data set provided by Michigan Peer Review Organization, Inc. (MPRO), each patient's risk-adjusted LOS is then linked to a quality-of-care judgment (good care, poor care) from physician peer reviewers. LOS is a widely used indicator of hospital performance. Most commonly, it is viewed as an indicator of hospital efficiency and as a surrogate measure for costs, with hospitals having long average LOSs considered relatively inefficient in the use of resources and those with low LOSs considered to be efficient. Sometimes, however, LOS is assumed to relate to quality. For example, if hospitals were to respond to the financial incentives of prospective payment by attempting to lower costs by prematurely discharging patients, LOSs significantly lower than expected might be considered indicative of poor quality care. On the other hand, if poor quality of care causes complications, it would tend to extend LOSs. Under this assumption, longer than expected LOSs could be viewed as indicative of poor quality care. This article shows that in every one of the 13 clinical conditions examined, cases that received poor quality care had significantly longer risk-adjusted LOSs than cases whose care was of acceptable quality.