Methacholine tests were used in an epidemiologic study of the prevalence of asthma and chronic bronchitis in northern Sweden. Of 6610 subjects in three age groups from eight representative geographic areas in the northernmost province of Sweden, 5698 (86%) completed a postal questionnaire on respiratory symptoms, and 1506 underwent a structured interview and a lung function test. A total of 292 (5%) were diagnosed as having asthma. A subsample of 284 subjects (of 320 invited) classified at the interview as having asthma (n = 98) or as having respiratory symptoms that might be due to asthma but not fulfilling the interview criteria for the diagnosis of asthma (n = 186) underwent a methacholine test. Subjects who, before the interview study, already had a well-defined asthma diagnosis were not invited to the methacholine testing. Of those 98 subjects classified as having asthma, 61% reacted to methacholine doses < or = 4 mg/ml and 79% to doses < or = 8 mg/ml, while the corresponding figures in the symptomatic but nonasthma group were 20% and 34%, respectively. The results show that a carefully performed structured interview accurately diagnoses asthma in epidemiologic studies. The methacholine tests provide important diagnostic information primarily in subjects in whom the medical history is equivocal.