This article presents latent growth modeling, a particular application of multilevel modeling, to examine the development of adolescents' math- and English-related talent perceptions and intrinsic values which are emphasized by Expectancy-Value theory as important precursors to a range of achievement-related outcomes. The longitudinal cohort-sequential study included participants in 3 overlapping cohorts, together spanning Grades 7 to 11 (N = 1,323). In this paper, I focus on the application of latent growth curve modeling to the accelerated longitudinal design and the interpretation of the growth parameters and their correlates in evaluating students' developmental trajectories across 2 key academic domains. Talent perceptions and intrinsic values were found to decline through adolescence in both math and English, and gender differences favored boys for math and girls for English consistent with gender stereotypes. (Contains 5 figures, 3 tables, and 3 notes.)