The relationship between airway hyperresponsiveness and longitudinal change in lung function was assessed in a population-based sample of 184 children and young adults observed over a maximum span of 12 yr. Pulmonary function was assessed annually with spirometry, and health and household information was obtained with standardized questionnaires. Nonspecific airway responsiveness to eucapneic hyperventilation with subfreezing air was measured on at least two occasions between the sixth and twelfth annual surveys. At any given survey, a significant bronchoconstrictor response was defined as [( prechallenge FEV1-postchallenge FEV1]/pre-FEV1) greater than or equal to 0.13, a value that identified 10% of the population. Subjects were classified as "never", "always", or "inconsistent" responders according to the consistency of responsiveness determined in different surveys. Subjects were classified further as "labile" if their maximal survey-to-survey difference in delta FEV1/FEV1 was greater than or equal to 0.18, and as "nonlabile" otherwise. A Markov-type autoregressive model that adjusts for previous pulmonary function level, sex, growth variables, and smoking exposures was used to model growth of FEV1, FEF25-75, and FVC. Overall, 135 (73%) of subjects never responded to the cold air challenge, six (3%) always responded, and 43 (24%) responded on one but not all occasions. Levels and rates of increase in level of FEF25-75 were significantly lower in the inconsistent and always responders. In contrast, levels of FVC were greatest and increased most in the always responders. In labile subjects, both FEF25-75 and FEV1 growth rates were reduced. These effects persisted when asthmatics were excluded from the analyses.(ABSTRACT TRUNCATED AT 250 WORDS)