Imago Dei has an interesting series of posts on the matter of whether Intelligent Design is a scientific or a metaphysical theory. The difference hinges on whether or not ID is falsifiable.
A theory is falsifiable, in the philosophical sense, if there is some conceivable set of observations which, if they occurred, would disconfirm the theory. If there is no imaginable observation which would ever count against the theory, then it is not falsifiable. This doesn't mean that one would ever make the observation, only that such an observation could be imagined. For example, the claim that there are an infinite number of other universes is unfalsifiable since there's no conceivable observation we could make which would prove it wrong. It may be true or it may be false, but it's not a scientific claim. Falsifiability is one of the criteria which distinguish scientific assertions from metaphysical claims.
As Imago Dei points out, ID's critics must think it falsifiable because many of them are trying hard to convince us that it's false. I had an amusing experience with this myself not long ago.
ID is falsifiable at least to the same extent as is the Darwinian assertion that natural selection is an unguided, purposeless process that is totally efficacious in itself to produce the grand diversity of life. Thus, if the latter claim is legitimately permitted in high school science classes, for example, then ID should likewise be permitted.
In a recent column for the local paper in which I supported the idea that ID was a legitimate topic for high school science classes, I made the point that if it could be demonstrated that a molecular machine like Michael Behe's bacterial flagellum could be shown to be constructed through mechanical means without intelligent input then, although ID might not be falsified in the strict sense (since it could still be argued that God had designed the mechanisms that led to the development of the structure), it would certainly be de facto discredited.
Someone wrote in to the paper and stated in rather snarky accents that in fact this had already been done and cited the web site to prove it.
Well, the writer's grandiose assurances were not justified by the information at the web site that he had commended to the paper's readers, but that wasn't the most important aspect of his reply. People who oppose having ID taught in science classes often do so on the premise that it's not science, and they buttress the opinion that it's not science by claiming that it's not testable, i.e. not falsifiable. My critic didn't seem to be aware that by asserting (wrongly) that a major ID claim had been shown to be false he was tacitly acknowledging that ID meets a crucial criterion of scientific theories. Without realizing it he was sawing off the branch upon which he sat.
Critics of ID are in a tough spot. They can't allege that ID is false without admitting that it's falsifiable, and they can't acknowledge that it's falsifiable without admitting that it's a valid topic for a science class.
The links at Imago Dei have much more on this controversy.