The genetic variability underlying many morphological and stress resistance traits may largely depend on the effects of genetic drift balanced by polygenic mutation. This model of adaptive potential has played a central role in the minimum viable population size concept and has been used to predict the effective population size necessary to prevent extinction within changing environments. However, there have been few long-term experimental studies of adaptive potential within isolated populations, and no study has thus far provided an experimental test of the drift-mutation model of quantitative genetic variation. Using the sternopleural bristle number of Drosophila melanogaster as a model quantitative trait, we performed repeated measurements of adaptive potential on 15 replicate populations of two and 10 male-female pairs over 30 and 77 generations, respectively. Declines in adaptive potential were analyzed by comparing observed and expected changes in realized heritability over time. The only significant model deviation occurred immediately after bottlenecks of two pairs, in which greater than expected declines in realized heritability were observed. This result suggests that changes in allelic diversity during bottleneck events may be as important as changes in heterozygosity in determining adaptive potential. Drift-mutation model expectations were otherwise realized over all generations. Our results validate the use of the drift-mutation model as a tool for understanding the dynamics of adaptive potential for peripheral fitness characters, but suggest caution in applying this model to recently bottlenecked populations.