In underregulated food systems, human mycotoxin exposure is common and causes adverse health outcomes. Maize is a staple crop in many parts of sub-Saharan Africa, and it is frequently contaminated by mycotoxins such as aflatoxin and fumonisin. Mitigating the effects of mycotoxins is challenging, as a diverse set of factors contribute to the overall mycotoxin burden. To address how monitoring could be implemented within underregulated systems, I first proposed a systems-level framework that includes responses for acute and chronic mycotoxin risk, using low-cost technologies to expand monitoring, and collaboration among regional monitors and local stakeholders. Many ideas in the proposed framework were tested by surveying a maize smallholder system in central Tanzania. Surveys of local grain mills enabled modelling of aflatoxins and fumonisins as a function of environmental, survey, and spectral data. These models demonstrated the ability of all covariate types to explain variation in mycotoxins and suggested that models can be improved by integrating data on diverse risk factors. In the face of resource constraints, accessible technologies are needed to mitigate human mycotoxin exposure. I tested a low-cost sorting device that separated maize kernels based on their physical properties. This device was validated with plastic kernel models of varying densities and then was used to reduce fumonisin in contaminated maize. From a plant breeding perspective, there are potential public health dangers in cases where linkage or pleiotropy results in an increase of yield and decrease in disease resistance. This work identified numerous significant loci controlling various ear traits, and some of these loci were also associated with effects on Fusarium ear rot (FER) traits. A variety of interactions were observed between ear and disease traits, indicating that increased ear size does not always increase susceptibility to FER and fumonisins. Some loci exhibited yield-mycotoxin tradeoffs, so caution should still be used, especially by breeders in mycotoxin-prone environments. In total, this dissertation demonstrates significant progress in several aspects of addressing the multifaceted challenge of mycotoxin contamination, especially in underregulated food systems. By taking a multi-disciplinary approach that addresses multiple scales, it can help to inform efforts in reducing human mycotoxin exposure.