In a longitudinal population study, 855 men, born in 1913 and initially examined when 50 years old, were followed for 17 years with measurements of dyspnoea and other variables performed at ages 50, 54, and 67 years. In addition a sample of 226 men born in 1923 was followed from 50 to 57 years of age. At the latest examination, four different methods for measuring dyspnoea were used, one based on questionnaire, one on interview, and two on visual analogue scales. The estimates from these methods were highly intercorrelated, and correlated with measures of cardiopulmonary function as well. The prevalence of dyspnoea grade 2 (shortness of breath when walking with someone of the same age on the level) or more, not counting the mildest form of dyspnoea in these populations, was 2.8%, 3.0%, 5.2% and 10.3% at 50, 54, 57 and 67 years of age, respectively. Dyspnoea grade 1 (shortness of breath when walking quickly on the level or uphill) was less well related to age. A scoring system to differentiate various possible causes of dyspnoea was applied. About one third of the dyspnoeic men had signs and symptoms of cardiac disease, one quarter had pulmonary disease, and a quarter had a combination of both causes. The remaining 20% had no signs or symptoms indicating cardiopulmonary disease but in the majority of the cases other plausible causes were found.