In this article, an innovative approach to improving the cyber-physical security of energy-efficient powertrain system in a hybrid electric vehicle (HEV) against sophisticated cyberattacks is presented. To the best of our knowledge, cyberattacks, especially sophisticated and subtle cyberattacks, have not yet been studied in energy management systems (EMSs) for HEVs. First of all, we present a systemic assessment of long-term sophisticated cyberattacks that aim to deteriorate the battery lifetime and energy efficiency of HEVs. Specifically, three levels of attack taxonomy according to the skill level of the attackers are considered, which are sophisticated and can hardly be detected by the human driver. In addition to levels 1 and 2, cyberattacks that do not or partially require prior knowledge of the vehicle, and we explore two other types of level 3 damage-oriented controller attacks made by highly skilled attackers who have sufficient knowledge of the system. Such sophisticated attacks will potentially cause severe damages, such as decreasing battery capacity and energy by up to 50%. For a comprehensive vulnerability assessment, we propose innovative evaluation metrics to analyze the impact and stealthiness of sophisticated attacks. Finally, a preliminary probability-based detection method for sophisticated damage-oriented controller attacks is developed to improve the cyber-physical security of energy-efficient powertrain system in HEVs.