OBJECTIVE--To assess attendance at and the characteristics of patients attending health checks for cardiovascular disease offered in a general practice over a period of five years (1984-9). DESIGN--Medical record audit and postal questionnaire survey. SETTING--One general practice in Oxfordshire with a socially diverse population. PARTICIPANTS--1101 Men and 1110 women aged 35-64 registered with the practice. MAIN OUTCOME MEASURES--Age, sex, marital state, social class, smoking habits, alcohol consumption, and diet. RESULTS--Of the 2211 men and women in the target age group (35-64) in 1989, 1458 (65.9%) had been offered screening and 963 (43.6%) had attended for a health check. Attenders were more likely to be women, aged greater than or equal to 45, married, non-smokers, and of higher social class than patients who did not respond to the invitation. The relative likelihood of non-attendance was 1.24 for smokers, 1.20 for the overweight, 1.16 for heavy drinkers, and 1.28 for those with a less healthy diet, even after adjustment for age, sex, marital state, and social class. CONCLUSIONS--After five years of offering health checks, opportunistically (to men) and in the context of cervical smear tests (to women), less than half of the eligible patients had attended. The likelihood of acceptance of an invitation to attend was inversely related to the patient's cardiovascular risk for all factors measured except age. A coherent strategy to reduce cardiovascular disease depends on more careful targeting of scarce health service resources and more emphasis on public health measures (such as dietary regulation and tobacco taxation). Doctors should be careful not to absolve the government of its public health obligations by substituting unproved preventive interventions aimed at the individual patient.