David Hume provided a compelling argument against believing in reports of miracles in chapter 10 of his Enquiry Concerning Human Understanding. The basic idea is this: we have to weigh the likelihood of a miracle taking place against the possibility that the report is exaggerated or simply false. The end result is that it is always more likely that the report is somehow wrong than that a miracle took place. For we have lots of experience of witnesses getting things wrong, or stories getting garbled as they are passed along, and really no solid firsthand evidence of miracles taking place. To believe such a report, you have to say, “The falsity of this report would be an even greater miracle than the one described in the report,” and that just never happens.
I think this argument can be extended to religions themselves. Every religion is based upon or implies claims that are supernatural — they would be judged as miraculously true, from the perspective of natural science. God speaking to humans, life after death, resurrection, rebirth, magical healing — all of these things, at least if taken literally, simply can’t be accommodated by natural science. (Or, if they can be naturally explained, then they lose their religious significance.) So in these cases, one should weigh the likelihood of these claims being true against the possibility that humans are being superstitious, or are being led by acritical wishful thinking, or are simply being taught to believe false things. And these phenomena are all too common in ordinary human experience. So, with any supernatural claim, it is always more likely that humans have somehow deluded themselves than that the claim is true.