In two cohorts of boys (n = 167) and two cohorts of girls (n = 131), FEV1, stature, body weight and two thoracic dimensions were measured eight times at intervals of about six months; the mean ages of each cohort at the start of the study were 12.5 and 13.5 years respectively. In each of the repeated cross-sections, FEV1 could be described as a power function of stature (FEV1 = aHk). In boys, the power (derived from regression of In FEV1 on ln H) increased from 2.60 to 3.04 and declined to 2.40; in girls, it decreased steadily from 2.87 to 1.94, attaining adult values at about age 15. A similar but less pronounced trend was observed for ln FEV1 against ln body mass. The adolescent growth spurt could be best studied in the boys. The individual (longitudinal) log-log plots of FEV1 versus stature or thoracic height were obviously non-linear in 52% and 33% of the boys respectively, with the greatest increase in FEV1 towards the end of the curve. In 78 boys with an apparently straight ln FEV1-ln H plot, the slopes of individual curves varied between 1.86 and 7.53; the mean value (3.99) was larger than in any of the repeated cross-sections. Applying allometric principles of similarity, FEV1 was about isometrically related to body mass but varied as about the 3.5th power of stature, with age-related trends in boys and girls. These findings reflect changes in bodily proportions and shape during the adolescent growth spurt, with growth in lung volume lagging behind growth in standing height.(ABSTRACT TRUNCATED AT 250 WORDS)