Recent advances in energy density of Li-ion cells together with high current fast-charging ask for improved strategies for on board safety and reliability observation of the cells. Potential degradation effects are stimulated by lithium plating and dendrite growth. The latter may ultimately cause an internal short circuit of the cell and can lead to serious damage. Increased self-discharge is an early indicator for safety-critical cell conditions. In this work, a method to calculate the self-discharge of Li-ion cells is presented. It is based on the analysis of cell voltage gradients during idle periods, which can be applied in state of the art battery management systems with low drift cell measurement. However, transition into the idle state after driving requires a settling time of several hours before the voltage gradient can be determined unambiguously. For the new accelerated self-discharge determination, a model-based approach is developed, which also considers aging effects of the open circuit voltage and the cell capacity. The self-discharge behavior of more than 100 cells is studied using the presented method during a 48-week aging experiment with real driving current profiles. Detailed information about the impact of various cycling and aging parameters on self-discharge is presented, including the influence of temperature, charge throughput, profile peak current, and state-of-charge range as well as aging of capacity and internal resistance. The results allow for the definition of optimized operation strategies for reduced safety risk, increased life time and determination of early warning criteria.