What are the prospects for meaningful learning in high-level creativity? An expertise acquisition view of creativity is optimistic about longitudinal improvement; a "Darwinian" view, emphasizing chance, is not. To address this issue, the complete outputs of 65 eminent composers (15,657 works) were tabulated and weighted; each work's masterpiece status was operationalized by citation and recording count measures. Works were assigned to 5-year age intervals, synchronized across composers by age at 1st masterwork. Hit ratio (masterwork-level music divided by total output) was calculated for each interval for each composer. The overall hit ratio trajectory was a single-peaked function, with a maximum around age 50. Reliable age effects held even during composers' mature periods, inconsistent with the chance view and more consistent-although not entirely so-with the expertise acquisition view. Several variables predicted individual differences in hit ratio trajectories; most notably, composers who wrote their best works later in their careers showed stronger agewise increases in hit ratio during their mature periods, even when best works were removed from the trajectories. Implications and prospects for reconciling the expertise and chance views, via a typology of conceptualist versus experimentalist creators, are discussed.