Libertarian theories of freedom and responsibility face a worry about luck: if an action is undetermined, the action cannot be legitimately attributed to the agent; instead the action is a matter of luck, and so the agent is not responsible for the action. Robert Kane defends libertarianism by appealing to self-forming actions. These actions are undetermined because the agent is attempting to act on two conflicting motives, but the agent is responsible for the outcome if she is responsible for having (...) those motives. If the agent “endorsed” both motives in earlier SFAs, Kane argues that we can hold the agent responsible for both motives, and hence responsible for the later SFA. We will further develop others’ arguments that Kane’s appeal to earlier SFAs to explain responsibility for a later SFA is unsatisfactory. We then raise a second objection to Kane’s use of SFAs. On one formulation of an SFA, the agent does three things: she exerts two efforts of will to make opposing choices, and she also acts on one of them. On another formulation of an SFA, the agent does just two things: she exerts two efforts of will to make opposing choices. One of those efforts turns out to be successful, but that success doesn’t require any further intervention from the agent. We distinguish two worries involving luck, a problem of resultant luck and a problem of constitutive luck, and show that neither model of SFAs solves both problems. (shrink)