It has been suggested that young persons who smoke have better lung function initially than those who remain nonsmokers. To examine this possibility prospectively, we analyzed respiratory questionnaire responses and lung function results in residents of Burbank and Lancaster, California, who had completed field screening studies of respiratory status at 2 times 5 yr apart. At Time 1 and Time 2, we calculated the age- and height-adjusted values for forced vital capacity (FVC) and forced expiratory volume in one second (FEV) of all white residents who at Time 1 were 13 to 23 yr of age and did not smoke tobacco. Dividing these into 2 groups, "starters" and "nonsmokers," we used analysis of covariance for males and females with height and age as covariates and compared lung function values at Times 1 and 2 and changes in lung function between these times. Among males at Time 1, FVC, FEV, peak expiratory flow, and maximal flow after exhalation of 25% of FVC were significantly larger for starters than for nonsmokers. At Time 2, values for these same indexes (except for FVC) were no longer significantly different between starters and nonsmokers. Our findings suggest that (1) relatively poor lung function may discourage young males (but not young females) from becoming regular tobacco smokers; (2) prediction equations based on so-called normal populations of nonsmokers might underestimate normal lung function, and (3) the adverse effect of smoking on lung function may be even greater than that estimated from cross-sectional studies.