In 1991, we began a project to search for the genetic basis of asthma using linkage analysis. We encountered discord between a history of asthma and physiological measures of variable airflow obstruction and sought to examine the frequency of such occurrences and the issues surrounding phenotyping of patients with asthma. We reviewed our experience in ascertaining the asthma phenotype in 50 nuclear families comprised of 219 subjects (110 male, 109 female). Three respiratory physicians reviewed data including a questionnaire, skin testing, objective measures of variable airflow obstruction [increase in FEV1 > or = 15% following salbutamol 400 micrograms of PC20 (methacholine) < or = 4 mg/ml], and serum for IgE. Thirty-eight percent of subjects had both objective and questionnaire data consistent with asthma (++) whereas 39% had negative objective and negative questionnaire findings (--) (i.e., no asthma). A positive history but negative objective findings occurred in 7% of subjects, 2% had a negative history and positive objective findings. Retesting was requested in 13% of subjects; review of historical data was requested in 1% (i.e., childhood asthma but no present asthma). Retesting was requested for either (a) positive history, negative objective if symptoms were seasonal or the subject was using medications known to affect the challenge study, (b) viral infection within 6 weeks of a positive methacholine study, or (c) technically inadequate study. Overall, after the initial assessment, all members of only 22 families could be catagorized as either ++ or --. The diagnostic group requested at least 1 retest in 19 families and a review of historical records in 2 families. We conclude that discordance between self-reported questionnaire data and laboratory measures of variable airflow limitation is common and will increase the numbers of asthmatic subjects in studies that seek to determine the genetic basis of asthma.