This essay evaluates John Martin Fischer and Mark Ravizza's mature semi-compatibilist account of moral responsibility, focusingon their new theory of moderate reasons-responsiveness as a model of "moral sanity." This theory, presented in _Responsibility and Control_, solves many of the problems with Fischer's earlier weak reasons-responsiveness model, such as its unwanted implication that agents who are only erratically responsive to bizarre reasons can be responsible for their acts. But I argue that the new model still faces several problems. It does not (...) allow sufficiently for non-psychotic agents (who are largely reasons-responsive) with localized beliefs and desires incompatible with full responsibility. Nor does it take into account that practical "fragmentation of the self" over time may also reduce competence, since moral sanity requires some minimum level of narrative unity in our plans and projects. Finally, I argue that actual-sequence accounts cannot adequately explain sane but weak-willed agency. (shrink)