Abstract : This thesis investigated the optimization process of composite plate girder bridges to AASHTO LRFD provisions using the evolutionary optimization model within Excel Solver. Evolutionary [also known as genetic] algorithms implement principles of natural selection and survival of the fittest in a mathematical model that seeks to find the best solution to a problem. Commercially-available software tools such as Solver, have streamlined the process of optimization to a degree where individuals can directly implement a form of optimization to a problem or product in order to gain market leverage and reduce costs, without any programming perquisites. The design process of composite plate girder bridges is a highly iterative procedure that can be swayed in many unpredictable directions, given the site conditions. As a result, generalizing a set of [golden]-rules, doesn’t always yield a commercially optimum design. Moreover, at the heart of AASHTO LRFD provisions, you will find yourself confined in a discrete, highly nonlinear problem, where conventional methods do not scale well. This is where the use of metaheuristic algorithms shines. The implemented genetic algorithm facilitated the search for optimum design configurations for a single-span composite plate girder bridge, having 9 variables. While also adhering to 57 constraints. The design constraints covered a wide spectrum of the code provisions, ranging from Strength, Fatigue, Service limit states to constructability and wind checks. 3 different models have been established and results were validated with existing literature. 20.5% weight reduction in the steel component was achieved, presenting a 2.1% improvement on referenced literature while also reducing the computation time to 5% of the referenced example. C The model confirmed site conditions permitting the use of deeper webs will yield the highest return on investment in terms of capturing the maximum protentional of the girders for the least amount of material, as opposed to adding additional girders. Finally, the viability of Solver as a readily-available tool for addressing the design of highly complex and nonlinear structures was confirmed.