Genome-scale metabolic models are powerful tools in helping to understand the metabolism of living organisms. They can be applied to the biomedical and biotechnological arenas. The use of such models enables fundamental understanding of metabolism and identification of drug targets in pathogenic microorganisms. They also facilitate metabolic engineering of recombinant organisms to make products useful to society. These mathematical models of metabolism are created based upon the genome annotation of the organism of interest. However, development of high quality versions of these models is non-trivial due to incomplete knowledge regarding gene function, as well as errors in genome annotations. Models developed under such circumstances display "metabolic inconsistency" and are mathematically infeasible. Genetic algorithms may be used to resolve these inconsistencies. In the process, it is possible to take advantage of the ensemble information inherent to the evolving population to gain additional biologically relevant insight. Specifically, it is possible to identify the most pathologic metabolic inconsistencies in an organism, facilitating experimental design and hypothesis development.