Two extreme theses can be formulated with respect to analogies: 1. Analogies have a temporary and a purely heuristic function. 2. All our thinking is conducted in terms of analogies. The first thesis has its natural locus in 1950s and 1960s philosophy of science (and before), the second in cognitive science and the mental models literature. Both these theses are wrong in their own way, although analogies do play a significant role in scientific reasoning. One difficulty is that there are a fair number of different candidates for the role of analogies in science. There is, for instance, a peripheral use of analogy merely to illustrate points, and there are other cases where analogies are crucial to the development of a scientific account. While analogies are sometimes the path to new insight, they are at other times a tool for the continuing thought about a scientific subject. I shall illustrate these different uses with examples and then review the methods of examining analogy use in science. My conclusion with regard to the above theses is located somewhere in the middle between the two extremes. While it is impossible to imagine, in practical terms, how science could survive without analogies, it is also not possible to make the argument that science requires analogies.