The role of microbial secondary metabolites in the ecology of the organisms that produce them remains poorly understood. Variation in aflatoxin production by Aspergillus flavus is maintained by balancing selection, but the ecological function and impact on fungal fitness of this compound are unknown. I hypothesize that balancing selection for aflatoxin production in A. flavus is driven by interaction with insects and microbes. To test this, I competed naturally occurring aflatoxigenic and non-aflatoxigenic fungal isolates against drosophila larvae and soil microbes using microcosm experiments. In all microcosms I used quantitative PCR to quantify DNA as a proxy for fitness. I demonstrate that, in the presence of insects, aflatoxigenic isolates have a fitness advantage relative to non-aflatoxigenic isolates. I speculate that this advantage is conferred through interference competition and protection from fungivory. Conversely, I demonstrate that aflatoxigenic isolates have a fitness cost relative to non-aflatoxigenic isolates when competing with soil microbes. I speculate that this fitness cost is the result of energetic costs associated with aflatoxin production. Using field isolates from two north-south transects spanning the United States, I demonstrate that patterns in the frequency of aflatoxigenic isolates are consistent with selection at a local level. To the best of my knowledge, this work represents the first evidence for possible selective pressures driving balancing selection and the most comprehensive assessment of A. flavus population structure in the United States to date.