An increasing proportion of the population is not currently employed; health inequalities research should include these people in analyses. Labor market position is the key determinant of health in Western societies; it is essential to clarify the independent effects of the three main components of labor market position: educational qualifications, occupational class, and employment status. These issues are addressed using national data from the British General Household Survey for 1991-1992 and two measures of ill-health: limiting long-standing illness and self-reported health. Occupational class is a stronger predictor than educational qualifications of measures of ill-health for men and long-term illness for women. When employment status is included in the models, the class gradient is weakened, suggesting that omitting the nonemployed from class analyses will produce a seeming weakening of class inequalities in health. The analyses show that class inequalities in health are somewhat greater for nonemployed than employed men, but weaker for nonemployed than employed women. In spite of older people having left the labor market many years earlier, class based on their previous occupation continues to have a major effect on their health. Class should be the primary variable in analyses of inequalities in health, and all groups of the nonemployed should be analyzed using their last main occupation.