Many claim that the “Trolley problem” is hindering the introduction of self-driving cars. Self-driving cars must, as all safety-critical products, be designed such that the probability of morally hard (“trolley”) situations is acceptably low. In this paper we argue that the introduction of self-driving cars can solve this ethical dilemma. In short the solution to the trolley problem is that a self-driving car must be able to estimate its own operational capability for handling surprising situations, and adjust its own tactical behavior accordingly. By limiting the risk for the case of not being able to handle all surprising events in a similar way as for other safety goals today, the remaining risk for the trolley problem can be argued as low as any other acceptable risk of vehicle E/E implemented functionality.