OBJECTIVE To explore the relationship between self reported environmental tobacco smoke exposure (or passive smoking), the serum cotinine concentration, and evidence of respiratory or coronary disease in men and women who have never smoked. METHODS Cross sectional random population survey identifying disease markers and relating them to measures of passive smoking. Disease markers were previous medical diagnoses, response to standard symptom questionnaires, and electrocardiographic signs. METHODS Samples of men and women aged 40-59 years drawn from general practitioner lists in 22 local government districts of Scotland, between 1984 and 1986. METHODS A total of 786 men and 1492 women who reported never having smoked tobacco, and who had serum cotinine concentrations below 17.5 ng/ml, the cut off point for smoking "deceivers", took part. RESULTS Fewer than one third of never smokers reported no recent exposure to environmental tobacco smoke and the same proportion had no detectable cotinine. Women had lower cotinine values than men but reported more exposure to smoke. The correlation between the measures of exposure was poor. Self-reported exposure showed strong, statistically significant, dose response relationships with respiratory symptoms and with the coronary disease markers. These relationships were weak or absent for serum cotinine, except for diagnosed coronary heart disease. Here the dose response gradient was as strong as that for self report, with an odds ratio of 2.7 (95% CI 1.3, 5.6) for the highest v the lowest exposure group, adjusted for age, housing tenure, total cholesterol, and blood pressure, and not explained by fibrinogen. CONCLUSIONS The validity of different measures of tobacco smoke exposure needs further investigation. The gradient of diagnosed coronary heart disease with both self reported exposure and serum cotinine was, however, surprisingly strong, statistically significant, and unexplained by other factors. These findings reinforce current policies to limit passive tobacco smoke exposure.