Systems biology is an emerging field of research that utilizes high-throughput experimental data and computational analysis methods to study biochemical networks. One of the common denominators of systems biology is the genome-scale reconstruction of metabolic networks. These reconstructions are biochemically, genetically, and genomically (BiGG) structured knowledgebases that seek to formally represent the known metabolic activities of an organism. In the first part of this dissertation, the general properties of these reconstructions, along with the use of constraint-based modeling to analyze these networks, is described in detail. Metabolic network reconstructions have many practical uses, including use in discovery of new gene functions and metabolic reactions, and for metabolic engineering. The core E. coli metabolic model, a small- scale model that can be used for in-depth analysis of new constraint-based methods, is presented and analyzed in detail. In the second part of the dissertation, the genome -scale metabolic reconstruction of E. coli was updated and analyzed. This reconstruction was first published in 2000, and has been updated and periodically published since then. The current version of the reconstruction is called iJO1366, and was updated based on new literature and database information. The remaining network gaps were analyzed, and a new gap-filling workflow was developed and used to predict the missing metabolic reactions and genes in iJO1366. Model predicted growth phenotypes were compared to a large experimental dataset of gene knockout strain phenotypes, and model errors were identified. Several in vivo experiments were performed to validate these model-based predictions. In the third part of the dissertation, the metabolic network model of E. coli was used for metabolic engineering. A large-scale computational screen using the algorithms OptKnock and OptGene was performed to design many growth-coupled production strains for various useful compounds. In order to validate the accuracy of these predictions and to determine if adaptive laboratory evolution can be utilized as an effective strain engineering tool, two strain designs were selected for experimental analysis. These knockout strains were constructed and evolved to optimize their phenotypes. One evolved strain worked as expected, producing a high yield of lactate from glucose. The other failed to produce 1,2-propanediol as predicted. This strain was analyzed by expression profiling, providing evidence for another metabolic gene function