The International Union Against Tuberculosis and Lung Disease questionnaire is widely used in epidemiologic studies of adult asthma. We examined whether the symptom questions could be classified into groups that represent different "syndromes," and whether some questions are better for predicting asthma than others. We analyzed questionnaire data from a population sample of 1,527 adults aged 18 to 55 years using factor analyses to classify the 17 respiratory symptom questions into four different groups that we termed asthma, cough, breathlessness, and urgent medical visit. The urgent medical visit was a subset of asthma. These four "syndromes" had good validity when measured against airway responsiveness to histamine, atopy to common allergens, lung function, smoking status, and body mass index. Questions that predicted asthma syndrome were those that asked about wheeze at rest or following exercise, asthma attack, chest tightness, and shortness of breath at rest. Questions about cough identified a different group of subjects who apparently did not have asthma. Questions of breathlessness did not aggregate with "asthma" or with "cough syndrome." The identification of particular questions that measure different respiratory conditions is important for epidemiologic studies when short questionnaires or more precise definitions are required.