Anne has been urging me to say something here about the somewhat counter-intuitive idea of "falsifiability": That scientific theory is truest that offers the most opportunities of being proved false.
Take, for example, string theory, the latest love affair of high-energy physicists and cosmologists. It's beautiful (I'll take the theorists word for it), and it offers the best hope yet of uniting the four known forces of nature, including gravitation. However, it makes no unique predictions that can be tested, at least not presently, and therefore no way of being invalidated. By the test of falsifiability, string theory doesn't qualify as science.
On the on the other hand, consider Einstein's 1915 theory of general relativity, which among other things uniquely predicted the deflection of starlight by the Sun, something tested by Eddington during the total solar eclipse of 1919. The observed deflection was exactly as predicted. According to the "falsification" criteria, the theory was good not because it was beautiful or because it had passed a dramatic test with flying colors, but because it had not muffed a prediction.
The falsification criteria was made almost a dogma of the philosophy of science by Karl Popper, in his 1934 book Logik der Forschung, published in English in 1959 as The Logic of Scientific Discovery. When my own interest in the history and philosophy of science began in the mid-1960s, Popper was all the rage. Making my way through Logic of Scientific Discoverywas one of the most unpleasant experiences of my life.
At the end, I suppose I had some illusion of understanding, but by the 1970s I was aware of a severe disconnect between what I was reading in the philosophy of science and what I was reading in the history of science. The former was invariably tidy; the latter, messy. Just now, as I'm writing, I went to the Q175s here in the college library and saw the shelves of books on the "logic" of scientific discovery that I had ordered and read as a young prof. It is, to be sure, a formidable bank of cogitation, but I wonder today what it has to do with how science works.
Lots of things go into a "successful" scientific theory -- including, perhaps, things that should be irrelevant, such as the prestige and personality of a theory's principle proponents. In the end, of course, nature bats last, and that's what gives us confidence in the process.
Still, if I had to pick a single philosophical criteria as most important in the actual practice of science, I'd say: That theory is best that explains the most in terms of the least. God shaves with Ockham's razor.