This paper examines factors that influence the veracity of verbal self-report data in health services research, using a cognitive social-psychological model of the data-gathering process as an organizing framework. It begins by briefly summarizing the consequences that can result from measurement error. Next, a cognitive social-psychological model of the question-answering process is presented. Common assumptions regarding the utility of specific assessment methods are evaluated with particular emphasis on the strengths and weaknesses of alternative data sources. The framework is then applied specifically to understanding the factors that may affect self-report measures in health services research relating to alcohol and other substance use. Overall, self-report procedures can provide useful estimates of consumption in clinical settings when conditions are designed to maximize response accuracy.