It might appear that the limits of toleration are so obvious that there is hardly a need to define them. Surely evil should not be tolerated. On the other hand, whatever is good should not only be tolerated but also encouraged. What is neither good nor evil should be tolerated; lest freedom of action and thought be impaired unnecessarily. Yet, this paper argues that these "obvious" truths are not obvious at all. The problem is that our understanding of the limits of toleration "presupposes" the distinction between good and evil, which raises two difficulties. The first is that there is considerable difference of opinion about what makes good dispositions, actions and policies good. Beyond that, there is a problem of applying the distinction between good and evil to particular cases, which we typically find to be good in some ways and not so good, or even evil, in other ways. Distinguishing good from evil presupposes a sophisticated form of critical self-knowledge that takes care not to assume that any description of an individual case can be readily extended to all the particulars of that case. Therefore, it paradoxically seems that there are times when intolerance should be tolerated; and times when toleration should not be tolerated. The paper argues that the way to a coherent view of toleration is to distinguish various degrees and levels of toleration. The analysis presented is especially indebted to the writings of John Locke, John Stuart Mill and Jean-François Lyotard.