Extensive literatures exist on the epistemology of testimony, memory, and perception, but for the most part these literatures do not systematically consider the extent of the analogies between the three epistemic sources. A number of the same problems reappear in all three literatures, however. Dealing simultaneously with all three sources and making a careful accounting of the analogies and disanalogies between them should therefore avoid unnecessary duplication of effort. Other than limits on the scope of which memorially- and testimonially-based beliefs should be included in the Parity Thesis, I argue that most of the disanalogies that different philosophers have proffered between the sources do not mark distinctions among the universes of possible testimonially-, memorially-, and perceptually-based beliefs regarding the explanation of those beliefs' epistemic status. I first criticize the suggestion that perception is a generative epistemic source, while testimony and memory are not; I propose and defend counterexamples in which testimony and memory produce new beliefs. Next, I criticize a variety of distinctions that have been drawn between testimony and perception, taken chiefly from the reductionist-antireductionist literature on testimony. I criticize the suggestion that the conceptualization of content and the transparency of experience affect the epistemologies of testimony and perception in different ways. Regarding memory and testimony, I advocate modeling testimony on the legal relationship of a principal and an agent, arguing that law's apparatus used to analyze such situations suggests that using others' epistemic services in testimony will supply the same epistemic benefits and burdens as if we had performed those epistemic tasks personally and then relied only on memory. I apply this analysis to the transmission of defeaters in testimony. I argue that memory does feature the epistemic equivalent of a perceptual image and that both perceptually- and memorially-based beliefs can concern either the past or the present. Finally, I construct a set of six transformations that turn individual possible instances of perceptually-, memorially-, or testimonially-based beliefs into individual possible instances of the other two types of beliefs without changing the structure of those beliefs' epistemologies.