Lithium-ion batteries (LIBs) are essential for the electrification of vehicles and play an important role for stationary storage units needed for grid-balancing. Research groups all over the world work on the improvement of LIBs regarding an increase in energy density as well as cycle-life and a decrease in costs. Next to these research topics, a continuously uprising and crucial field is safety features of LIBs, which can be implemented at different levels, such as material, cell, battery and system level.[1] The abuse/misuse of a LIB can cause an internal release of heat which can trigger a chain of exothermic reactions on cell level. Hence, the cell temperature increases dramatically, causing the so-called thermal runaway (TR), possibly leading to flames and/or explosion of the cell.[2] Moreover, the TR of one cell can initiate the TR of adjacent cells leading to a so-called propagation, possibly, causing the TR of the whole battery. Ideally, easily obtainable key features of a certain cell – such as cathode type, cell format, cell energy and state of charge (SOC) - could allow the prediction of its behaviour under abuse conditions. In the present study, we will discuss the latest result of our safety tests on cell level employing an external heater as TR-trigger. Single cell tests will be analysed regarding different hazardous features during the TR, e.g., cell temperature, occurrence of flames, peak pressure, gas amount and gas composition. Moreover, the possibility of a TR-propagation and the respective propagation speed will be gained from propagation tests utilizing six cells with identical SOC. In total, the study comprises over 200 tests on cell level. The gained data set is analysed in respect to the cell parameters, such as cell format, cell energy, SOC and the cathode type as well as the atmosphere (air vs. N2) present during the test. A special focus is put on the discussion of general conclusions linking cell parameters to TR-effects and propagation behaviour.The findings regarding common conclusions between key features and TR-effects can enable a rather facile selection process of cells/batteries for certain applications according to specific safety targets. Moreover, it allows to choose cell-specific safety measures, suitable during operation. In further works, the study will be extended to end-of-first life cells yielding important conclusions regarding crucial safety aspects for the implementation of those cells in 2nd-life application. Generally, the presented results can increase the overall understanding of the TR mechanism supporting the design of advanced measures to enhance the safety on cell level in the future.