The distribution of bronchial responsiveness (BR) and the place of symptomatic subjects in this distribution was investigated in a random population sample of 339 subjects who participated in the 1984 survey of the Vlagtwedde/Vlaardingen Study. BR was assessed by a histamine provocation test. The following indices of BR were used: (1) a histamine threshold both at 10 and at 20% decreases in FEV1 (PC10 and PC20); (2) the concentration causing a 10 or a 20% decrease estimated by linear interpolation (IPC10 and IPC20), and by linear regression of all data points (PD10 and PD20); (3) the slope of the linear regression line through all data points (SAP) and the slope of the line through the origin and the last data point (SOL). Disease was defined as the self-reported presence of respiratory symptoms from a respiratory symptom questionnaire. Bronchial challenge was considered as a test to predict symptom prevalence. Receiver-operator characteristic curves were applied to define optimal cutoff points for each of the indices used to define a positive and a negative test. For each index, sensitivity, specificity, predictive value of a positive and negative test, and efficiency rate of the test were calculated. The overall distribution of BR in this population appeared to be log normal for the different continuous indices. Subjects with symptoms were on the more responsive tail of the distribution. After adjusting for age, sex, smoking habit, and level of pulmonary function, the mean log-transformed values of PD10, PD20, SAP, and SOL for the various symptom groups differed significantly.(ABSTRACT TRUNCATED AT 250 WORDS)