With the rapid development of electric vehicle (EV), researches related to EV battery systems have flourished in recent years, partly because the battery system accounts for almost half of the cost of an EV. For the battery system, its capacity degraded gradually along with charge-and-discharge cycles and hence increases the risk of the EV. The reliability issue of the battery system thus arises. The present study proposes an analytical model based on reliability engineering for studying the capacity degradation and quantitative reliability of EV battery packs. The model, in turn, is constructed in consideration of states of health (SOH) of battery cells and their uncertainties. As compared to the traditional system reliability model that judges the pack’s reliability purely based on considering if or not a battery cell’s SOH exceeds a prescribed threshold, the proposed model adopts the k-out-of-n system reliability concept and considers that SOHs of battery cells within the pack do not differ too much to assure the pack’s reliability. This study also adjusts a few parametric values in the model to investigate their influences on battery pack’s life and reliability. Through numerical case studies, it is found the proposed model results in a more conservative system reliability prediction of a battery pack than the traditional model, and the capacity degradation trend of the battery pack coincides with a real case. In consideration of an effective battery management system that avoids fierce temperature changes, the influence of temperature is not very obvious in this study. The uneven performance of battery cells reflected by discrepancies of their SOHs is found crucial to the battery pack. If the performance of each battery cell is balanced, the battery pack’s life and reliability would increase significantly. The above trends coincide with those factors considered in a real battery management system (BMS) of an EV.