The aim of the study is to (i) identify common dietary patterns, (ii) study socioeconomic differences in these dietary patterns, and (iii) assess whether they contribute to socioeconomic differences in biological risk factors. The data come from the Whitehall II study of London civil servants, who participated in the third phase (1991-1993) and were 39-63-years old (N=8004). Food frequency questionnaire and socioeconomic background information was from a questionnaire, and biological risk factors from a medical screening. Six dietary patterns were identified. In reference to high employment grade men, the odds ratios of low grade men consuming the 'unhealthy' or the 'very unhealthy' diet were 1.26 and 3.34, respectively, while the odds for the 'French' diet was 0.13. Among women the corresponding odds were 2.98, 6.19 and 0.25. Adjusting for spouse's socioeconomic status and to a lesser extent smoking and exercise as well as job control attenuate these grade differences somewhat. Among men and women adjusting for dietary patterns accounted for about 25-50 per cent of grade differences in HDL and serum triglyceride levels.