Purpose: This research investigates what factors affect the degree to which nursing home performance explains variance in residents' change in status of activities of daily living (ADL) after admission. Design and Methods: The database included all residents admitted in 2002 to a 10% random sample of nursing homes in the United States. Longitudinal analyses of outcomes at 3 months after admission test the ability of individual characteristics and nursing home identifiers to explain variance in ADL change for different groups of residents. Results: As we compared the best and worst providers (top 20% vs bottom 20%, then 10%, then 5%) and we restricted analyses to more homogeneous groups of residents, we found that more of the variance in ADL change was attributable to provider performance. Cognitive function and race also affected the degree to which home performance had an impact on outcomes. Implications: The results imply that some quality indicators may be most useful in distinguishing between nursing homes that provide the best or the worst care. Furthermore, the degree to which a quality indicator is driven by a nursing home's performance may vary considerably, depending on the characteristics of the consumer. These findings raise questions about the usefulness of performance measures that focus on heterogeneous groups of consumers or entire provider populations. "How much of the variance in a quality indicator does provider performance explain?" is an issue we think has not received the attention it deserves in current discussions of performance-measurement strategies and pay-for-performance models.