The Whitehall study of British civil servants begun in 1967, showed a steep inverse association between social class, as assessed by grade of employment, and mortality from a wide range of diseases. Between 1985 and 1988 we investigated the degree and causes of the social gradient in morbidity in a new cohort of 10,314 civil servants (6900 men, 3414 women) aged 35-55 (the Whitehall II study). Participants were asked to answer a self-administered questionnaire and attend a screening examination. In the 20 years separating the two studies there has been no diminution in social class difference in morbidity: we found an inverse association between employment grade and prevalence of angina, electrocardiogram evidence of ischaemia, and symptoms of chronic bronchitis. Self-perceived health status and symptoms were worse in subjects in lower status jobs. There were clear employment-grade differences in health-risk behaviours including smoking, diet, and exercise, in economic circumstances, in possible effects of early-life environment as reflected by height, in social circumstances at work (eg, monotonous work characterised by low control and low satisfaction), and in social supports. Healthy behaviours should be encouraged across the whole of society; more attention should be paid to the social environments, job design, and the consequences of income inequality.