The trajectory design, propellant loading, and staging of a Mars Ascent Vehicle (MAV) can be accomplished via the solution of the associated optimal control problem under given vehicle, mission, path and control constraints. This paper presents the implementation of this optimization using Gauss Pseudospectral OPtimization Software (GPOPS-II) in combination with a Sparse Nonlinear OPTimizer (SNOPT) as the nonlinear programing solver. The solution obtained is an input to the propulsion, Guidance Navigation and Control, mechanical design, aerodynamic and thermal analysis teams, which use it to size systems, tune algorithms and update models. Optimal solutions are obtained for a variety of MAV architectures, which can be grouped in categories according to the number of stages (one or two) and the propulsion technology (solid, liquid or hybrid). A sensitivity analysis of variables of interest, like launch conditions, target orbit, stage and payload mass is presented.