This article sets out a formal procedure for determining the probability that God would do a specified action, using a subset of the theologian’s beliefs I will refer to as ‘moral knowledge’: our beliefs about what is right for us to do, and our beliefs about the details of situations which bear on what we should do. Why would we want to determine such probabilities? To begin answering, I point out how predictions of divine action are central to natural theology, as well as relevant to other pieces of religious reasoning (section 2). After setting out the procedure (sections 3-6), I will give a better answer to another obvious criticism. Perhaps natural theology has got by well enough so far with implicit or explicit judgments about what God would do, and when we have needed to be explicit about the probability of these predictions, ‘ball-parked’ figures based on ‘hunches’ or ‘intuitions’ have served us well, whereas the precision I offer brings needless complexity. I will therefore explain (section 7) how this increased precision helps interlocutors to structure discussions and focus on the deeper sources of disagreement in natural theological controversies.