So as fate would have it, I've written on the topic of how historians treat miracles quite recently here. This was in answer to a question about the miracles of Muhammed, but the principle is much the same. In fact, as a historian of late medieval (especially female) sainthood, I'm on much surer ground with Joan of Arc than I am with Muhammed. Indeed, my answer there was derived largely from a paper I'm working on about miracles in hagiographic sources, and how we as historians are to deal with them.

I'm going to expand on my earlier answer a little here (so I'd recommend having a look at that first), partly because I think the thrust of your question is a little different, and partly because your question gives us the opportunity to actually do a little real historical work, right here before your very eyes!

The main thing I'd like to expand on is that there's currently no real consensus on the best way for historians to tackle miracles. There's been some really fantastic research with miracle sources in the last 10-15 years, but most or all of that research has avoided tackling the issue of miracles head-on, and basically fallen back on Brian Stock's approach that I quoted in my above answer: 'the point is not whether the miracle “took place”, but that people whose social affiliations can be the object of empirical study explained their behaviour in terms of it'. Over the last few decades, historians have largely accepted this as a solid compromise that allows them to get on with the job of "doing history" (and quite reasonably so).

However, the work I'm doing at the moment requires a more nuanced understanding of miracles. I think Stock's formulation is accurate and useful, but I don't think it accounts for the fact that miracles were, to the medieval mind, very real indeed. It reduces the miracle to it's social impacts, to the point where the actual event is irrelevent - the only thing that matters (that makes the miracle 'real', if you like) is how people reacted to it. For Stock, the miracle 'exists' only so long as there is a quantifiable reaction it. To that end, I'm currently proposing an addition or qualification to Stock's statement: we are not trying to ascertain the empirical reality of the miracle, we are trying to consider the miraculous event as a tangible and textual manifestation of the underlying mental structures of our subjects as way of accessing those mental structures. I'm hoping (and this is ongoing work, so it's very much a hope) that this methodology will allow us to account for the fact that miracles were a key element in medieval epistemology, and not reduce them solely to social (or worse, socio-economic) impacts.