Inter-cell virtual connection is likely to occur in the process of electric vehicles driving, which could cause fire or explosion accident. This paper presents a connecting fault detection method of lithium-ion power batteries in series. The cross-voltage test is adopted to distinguish contact resistance increases and internal resistance increases fault. The battery voltage and negative surface temperature are collected by battery test system and auxiliary channels equipment. The experimental battery based on first-order resistance and capacitance (RC) equivalent circuit and MATLAB/Simulink platform is simulated. The mean square error which indicates the difference between experiment and simulation is employed to describe voltage state of the cell. If the abnormal voltage exists, it is concluded that a fault occurs through analyzing the voltage abnormal coefficients based on modified Z-score, which is considered as a second-degree fault. The temperature rise rate is regarded as a secondary parameter to determine whether the second-degree fault deteriorates into a first-degree fault. According to different levels of failure, battery management system reminds users to take appropriate measures in practice. This work can provide an effective method of detecting lithium-ion power batteries connecting failure in series. (C) 2018 Elsevier Ltd. All rights reserved.