Elite athletes and nonathletes (N = 1,268) attending the same selective sport high school (4 high school age cohorts, grades 7-10, mean ages varying from 10.9 to 14.1) completed the same physical self-concept instrument 4 times over a 2-year period (multiple waves). We introduce a latent cohort-sequence analysis that provides a stronger basis for assessing developmental stability/change than either cross-sectional (multicohort, single occasion) or longitudinal (single-cohort, multiple occasion) designs, allowing us to evaluate latent means across 10 waves spanning a 5-year period (grades 7-11), although each participant contributed data for only 4 waves, spanning 2 of the 5 years. Consistent with the frame-of-reference effects embodied in the big-fish-little-pond effect (BFLPE), physical self-concepts at the start of high school were much higher for elite athletes than for nonathlete classmates, but the differences declined over time so that by the end of high school there were no differences in the 2 groups. Gender differences in favor of males had a negative linear and quadratic trajectory over time, but the consistently smaller gender differences for athletes than for nonathletes did not vary with time.