Neural networking is a new approach to modeling batteries for electric vehicle applications. This modeling technique is much less complex than a first principles model but can consider more parameters than classic empirical modeling. Test data indicates that individual cell size, geometry, and operating conditions affect battery performance (energy density, power density and life). Given sufficient experimental data, system parameters, and operating conditions, a neural network model could be used to interpolate and perhaps even extrapolate battery performance under wide variety of operating conditions. As a result, the method could be a valuable design tool for electric vehicle battery design and application. This paper describes the on going modeling method at Texas A&M University and presents preliminary results of a tubular lead acid battery model. The ultimate goal of this modeling effort is to develop the values necessary for predicting performance for batteries as wide ranging as sodium sulfur to zinc bromine.