This report describes the social distribution of central obesity and the metabolic syndrome at the Whitehall II study phase 3 examination, and assesses the contribution of health related behaviours to their distribution. Cross-sectional analyses were conducted utilising data collected in 1991-1993 from 4978 men and 2035 women aged 39-63 years who completed an oral glucose tolerance test. There was an inverse social gradient in prevalence of the metabolic syndrome. The odds ratio (95% confidence interval) for having the metabolic syndrome comparing lowest with highest employment grade was: men 2.2 (1.6-2.9), women 2.8 (1.6-4.8). Odds ratios for occupying the top quintile of the following variables, comparing lowest with highest grade, were, for waist-hip ratio: men 2.2 (1.8-2.8), women 1.6 (1.1-2.4); post-load glucose: men 1.4 (1.1-1.8), women 1.8 (1.2-2.6); triglycerides: men 1.6 (1.2-2.0), women 2.2 (1.5-3.3); fibrinogen: men 1.7 (1.4-2.3), women 1.9 (1.2-2.8). Current smoking status, alcohol consumption and exercise level made a small contribution (men 11%, women 9%) to the inverse association between socioeconomic status and metabolic syndrome prevalence. In conclusion, central obesity, components of the metabolic syndrome and plasma fibrinogen are strongly and inversely associated with socioeconomic status. Our findings suggest the metabolic syndrome may contribute to the biological explanation of social inequalities in coronary risk. Health related behaviours appear to account for little of the social patterning of metabolic syndrome prevalence.