This paper presents details on the development of two mathematical models of lithium polymers batteries used in electric vehicles (EVs). These models describe the battery state of charge (SOC), and the output voltage of the batteries by using experimental data gathered during the driving of the EV on pre-established routes. The first model estimates the SOC based on temperature data, while the second model replaces the RC network of the traditional Thevenin model by a battery transfer function where the variables of temperature and current are used instead. To determine the performance of these models, simulations are conducted in the MATLAB/Simulink platform. Simulation results are compared with data measured from the EV by using the mean square error index (MSE) for each estimator. The second modeling approach shows a better performance than the battery empirical model obtained from Nernst model.