Abstract
In the first, shorter part of the paper I point out some problems and potential misunderstandings connected with B. Berčić’s treatment of Nozick’s sensitivity condition for knowledge. In the second part of the paper I offer the condition of modal stability or limited sensitivity as a revision of Nozickian conditions for non-accidental connection between our belief and the truth of our belief. “When it is seriously possible for you to falsely believe that p,” that is a good reason for denying that you know that p. Sensitivity within limits requires that we consider more possible worlds than classical sensitivity, but not the worlds outside this sphere. The idea of modal stability combines robustness with responsiveness to facts