BACKGROUND: The relationship between self-rated health and mortality after adjustment for sociodemographic variables, physician-rated comorbidities, disease severity, health-related quality of life (HRQOL), and psychosocial measures (depression, social support, and functional ability) was examined in the Mediators of Social Support (MOSS) study. SUBJECTS: The sample consisted of 2,885 individuals (mean age, 62.5 years) who had significant heart disease based upon heart catheterization. RESULTS. Using Cox proportional survival analysis, individuals who rated their health as "fair" or "poor" had a significantly greater likelihood of all-cause mortality (OR = 2.13; CI = 1.40-3.23; OR = 4.92; CI = 3.24-7.46, respectively) across follow-up (mean, 3.5 years) than those who rated their health as "very good" after considering sociodemographic factors. After adjustment for comorbidities, disease severity, HRQOL, psychosocial factors, and demographic variables, only those who rated their health as poor had a significant greater risk of mortality (OR = 2.96, CI = 1.80-4.85). A similar pattern was observed for coronary artery disease (CAD)-related mortality; increased adjustment of variables weakened the relationship between self-rated health and mortality. Individuals who rated their health as poor had a significantly greater risk of CAD-related mortality than did those who rated their health as very good (poor vs. very good OR = 3.58, CI = 2.13-6.02) after adjustment for all available mortality risk factors. CONCLUSIONS: This study indicates that it is important to include self-rated health when studying risk factors for mortality. Not adjusting for relevant factors may provide an overestimation of the effects of self-rated health on mortality in a sample of CAD patients.