A discussion of scientific knowledge requires paying special attention to the distinctive methods that science has developed to acquire and evaluate knowledge. There is no prima facie reason why, in the light of these methods, the epistemology of science should, to any great extent, resemble traditional epistemology, the study of knowledge gained by unassisted humans. Indeed, we shall see that not only are those traditional epistemological concerns often rather remote from those that are relevant to science, but insisting on applying constraints from older epistemological traditions can seriously distort our assessment of scientific knowledge.