A growing number of electric vehicles (EV's) are being used in fleet applications, creating a need for accurate estimates of vehicle mass while the vehicles are in operation. In this work, on-road energy use data are compared with simulated energy use to identify vehicle mass. The testing was performed on an electric Ford Transit Connect light-duty delivery vehicle in service with the Massachusetts Institute of Technology's facilities department. Driving data was collected using specific protocols designed to yield optimal inputs for identification, as well under normal driving conditions for evaluating the algorithms ability to identify parameters in worst-case scenarios. In this work, the identified mass is used to optimize fleet performance by providing more insight into the in-service weight of the vehicles, as well as by providing better electric vehicle range estimates to improve fleet utilization. Parameter identification methods have been developed for various other purposes, such as ensuring appropriate tire inflation levels, powertrain degradation and road quality monitoring, etc. but the methods and models presented in this work are deliberately simple to be useful in real-time applications. The methods presented here are designed for vehicles with well-conditioned efficiency maps such as EV's, Fuel Cell Electric Vehicles (FCEV's) and some hybrid architectures. While the models presented in this work are less complicated than many contemporary identification models, when combined with the accompanying signal analysis algorithms they are nevertheless able to consistently identify vehicle mass with less than 4% absolute error.