Objective: This study describes the associations between sedentary behaviour (television viewing) and participation in vigorous recreational activity with obesity and with biomarkers of cardiovascular disease (CVD) risk profile. Design: Cross-sectional analysis of the EPIC-Norfolk cohort study. Setting: The study is a population-based study of participants living in Norfolk, UK. Subjects: A total of 15 515 men and women aged between 45 and 74 y, recruited through General Practice lists, who completed the detailed physical activity questionnaire. Results: Following exclusion of those with self-reported myocardial infarction, stroke and diabetes, 14 189 participants remained for the analysis. Self-reported television viewing was positively and participation in vigorous activity negatively associated with markers of obesity, blood pressure and plasma lipids. In multiple regression analysis, adjusting for age, alcohol, smoking, treatment for hypertension, vigorous and total physical activity, these associations remained significant. For women who participated in more than 1 h/week of vigorous activity and who watched fewer than 2 h of television each day, the adjusted mean body mass index was 1.92 kg/m 2 less than for women who reported participating in no vigorous activity and who watched more than 4 h of television each day (Po0.001). The equivalent figure for men was 1.44 kg/m 2 (Po0.001). In a similar analysis, with blood pressure as the outcome, mean diastolic blood pressure difference between the extreme groups of vigorous activity and television viewing was 3.6 mmHg in men (Po0.001) and 2.7 mmHg (P ¼ 0.001) in women. Conclusions: These data suggest that time spent participating in vigorous recreational physical activity and television viewing, an indicator of a sedentary lifestyle, are associated with obesity and markers of CVD disease risk independent of total reported physical activity. Whether these observations represent the true underlying aetiological relations or are a manifestation of the different precision with which the subdimensions of activity are measured remains uncertain.