In A Theory of Conditionals, Robert Stalnaker distinguishes the ``logical problem'' of counterfactuals from the ``pragmatic problem'' of counterfactuals. Whereas the former problem has received much attention, the latter has received relatively little. This paper is concerned with a central aspect of the latter problem, namely the problem of change in the evaluation of counterfactuals (which corresponds to change in the selection function, under Stalnaker's theory). Firstly, it defends the importance of the problem: not only is it relevant in domains where counterfactuals play an important role (in epistemology, philosophy of science or decision theory, for example), but it can be relevant to philosophical debates concerning the nature of counterfactuals. Secondly, the paper attempts to lay the foundations of a framework for thinking about such changes, by distinguishing different sorts of change. These basic notions can be used to formulate and discuss principles concerning changes in the evaluation of counterfactuals. In the final part of the paper, a principle of this sort is proposed and defended. Its philosophical consequences are explored: in particular, it is shown how it can be used to rebut recent arguments by John Hawthorne and Alan Hajek which place traditional theories of counterfactuals in difficulty.