During 1978-80, 7735 men aged 40-59 took part in a screening examination which included an administered version of the WHO (Rose) questionnaire on chest pain. Those men who had 'possible' angina were as likely to have an ischaemic electrocardiogram at rest as those with 'definite' angina. Furthermore, the heart attack rate over 7.5 years was similar in men with 'definite' and 'possible' angina and did not diminish with length of follow-up. Five years after the initial screening 98% of the surviving men replied to a postal questionnaire which included a self-administered version of the chest pain questionnaire. Age-specific prevalence rates of angina based on the administered questionnaire at screening and on the self-administered postal version five years later, were similar. Men who had angina ('definite' or 'possible') on both occasions had the highest prevalence rate of ischaemic electrocardiograms, men who were positive on only one occasion had intermediate rates and those who were negative on both occasions had the lowest rates. We conclude that in cardiovascular studies of middle-aged men which use the WHO (Rose) questionnaire for the purposes of determining prevalence or assessing risk of a heart attack, angina should include both 'possible' and 'definite' angina.