Incidence studies offer a better opportunity to study risk factors for asthma than do prevalence studies. However, regular prospective follow-ups of large cohorts are difficult to perform, and that is why direct measurement of the incidence rate of asthma is almost impossible. Thus, cross-sectional follow-up studies of defined cohorts can be used to provide data on incidence. In 1986, a postal questionnaire survey on respiratory symptoms and diseases was performed in the northernmost province of Sweden. The population sample comprised all subjects born in 1919-20, 1934-5, and 1949-50 in eight representative areas of the province, which comprises 25% of the total area of Sweden. Completed answers were given by 5698 subjects (86%) of the 6610 subjects invited to the study. In 1992, the cohort was invited to a follow-up survey during the same season as in 1986, and 6215 subjects were traced. Of the 5393 subjects who answered the questionnaire, 4932 had participated in the 1986 survey, or 87% of those who participated in 1986. For the period 1986-92, the cumulative incidences of asthma were 4.9 and 5.0%, respectively, as assessed by the questions, "Have you ever had asthma?" and "Have you been diagnosed as having asthma by a physician?" Thus, the results indicate a mean annual cumulative incidence of asthma of 0.8%. After correction of the results for subjects who were diagnosed as having asthma in the clinical part later in the 1986 study, the mean annual cumulative incidence of asthma was found to be 0.5%. Risk factors were family history of asthma (OR 3.46) and current and former smoking, while female sex was a strong trend.