It is a common view that the confirmation of a hypothesis by experimental evidence is better if this evidence is novel. I think this common view is right, and my goal in this paper is to explain why this is so. The bulk of the paper is taken up addressing two important problems for the use of novelty considerations in science. These two problems result from what I term the 'relativity of novelty’, the fact that what counts as novel evidence is relative to a person’s background knowledge. I call these problems the ‘ignorance problem’ and the ‘biography problem’. According to the ignorance problem, given that ignorance generates proportionately more novel evidence (the less one knows, the more that is new), it appears that ignorance should be valued in science just to the extent that novelty is. According to the biography problem, since the novelty of evidence depends on a person’s state of knowledge, it follows that to evaluate the novelty of evidence and so assess its confirmatory significance we need to engage in various psycho-social investigations that reveal how much background knowledge a person possesses. But surely such investigations are foreign to the normal practice of science. My strategy in responding to these problems is to introduce a new interpretation of novelty (which I call ‘prima facie heuristic novelty’). After philosophically motivating this new version of novelty and illustrating its use through some historical examples, I argue that this new version of novelty provides an effective remedy to the ignorance and biography problems and thus explains the legitimate role of novelty considerations in science.