A practical universal modeling and simulation approach is presented in this paper to show that accurate battery pack simulation can be achieved if cell-to-cell variations were taken into account. A generic equivalent circuit model was used in the approach with parameters deduced from cell testing with proper protocols, which could come from live cell monitoring in a control circuitry. Using a single cell model, which was validated against experimental data and demonstrated with validity of high accuracy in predicting cell performance, we showed that such a high accuracy in single cell model is essential for a high fidelity pack simulation. It is also important to derive statistical confidence intervals accurately from experiments to characterize intrinsic cell-to-cell variations in capacity and internal resistance, which need to be considered in the pack model. If parameters for each individual cell were correctly approximated and used in the pack model, the accuracy in the prediction of pack performance could be significantly improved.