In this paper I discuss a central objection against diseases being natural kinds—namely, that diseases are processes or transitions and hence they should not be conceptualized in the ‘substantish’ framework of natural kinds. I indicate that the objection hinges on conceiving disease kinds as phase kinds, in contrast to the non-phase, natural kinds of the exact sciences. I focus on somatic diseases and argue, via a representative comparison, that if disease kinds are phase kinds, then exact science kinds are phase kinds as well. On the other hand, if exact science kinds are non-phase kinds, then disease kinds are non-phase kinds as well. This objection should thus be rejected, under a certain caveat, though. If natural kind membership has an influence over the diachronic identity of kind members, then it is possible, in principle, to draw the phase/non-phase distinction such that an ‘ontological gap’ lies between medical kinds and exact science kinds. I show further that this caveat is unavoidable even in relation to substantive universals and ‘essential’ properties—two controversial, strong features that were traditionally associated to natural kinds.