A priori metaphysics has come under repeated attack by naturalistic metaphysicians, who take their closer connection to the sciences to confer greater epistemic credentials on their theories. But it is hard to see how this can be so unless the problem of theory change that has for so long vexed philosophers of science can be addressed in the context of scientific metaphysics. This paper argues that canonical metaphysical claims, unlike their scientific counterparts, cannot meaningfully be regarded as ‘approximately true,’ and (...) that this means that the epistemic progress that science arguably enjoys through episodes of theory change cannot be expected to transfer to its metaphysics. What the value of engaging in metaphysics of science before the emergence of a final theory becomes correspondingly unclear. (shrink)