Philosophical theses are sometimes put by saying that one class of facts grounds another such class. Recently, several authors have tried to clarify and regiment such talk of grounding, and have asked about what features the relation of grounding has. The orthodox view is that grounding is a necessary connection among facts: if some facts ground another fact, they necessitate it, and the grounding fact itself is non-contingent. I shall challenge this view by presenting potential counterexamples. Finally, I will sketch a positive account of grounding that allows it to be contingent.