The fundamental assumption of cognitive psychology is that the cognitive mind is a representational system which mediates between sensory inputs and behavioral outputs. The primary task for the cognitive psychologist is one of explaining how the various cognitive capacities operate by reference to the structure of the salient parts of this representational system. The explanations offered are both functional and decompositional: they decompose the relevant capacities into their basic representational components and show how those components function together to produce the capacities. Theories are evaluated by how well they account for the behavior observed in psychological experiments and, at the lowest level, by how well they fit with knowledge gleaned from neurophysiology about the physical bases of the capacities. Philosophers have usually assumed that pain cannot lie within the domain of cognitive psychology. Pains, it has been supposed, are not like images or memories or visual percepts: they have no representational content. So, there can be no explanation of the desired sort. To understand the various facets of pain, we need to look elsewhere, perhaps to the realm of neurophysiology. Cognitive psychology cannot help us. This, I now believe, is, a mistake: pains do have representational content. So, the view that pain is not a proper object of study for cognitive psychology is not well founded. My discussion begins with an old objection to the token identity theory in connection with after-images, and a modem response to it which has become widely accepted. This response, I maintain, is unsatisfactory, as it stands. But, with one key revision, it is, I believe, defensible, and it has ramifications for our understanding of pain. In particular, it points to the conclusion that pains have representational content, as does at least one other facet of our everyday conception of pain. In the third section of the paper, I consider the question of what sorts of representations pains are most plausibly taken to be. Are they sentences in an inner language, like beliefs and desires, on the usual computational conception of the latter states? Or are they representations of a different sort? I suggest that a purely sentential approach is difficult to reconcile with some of the neuropsychological data on pain, and I make an alternative hybrid proposal. Pains, I propose, are representations of the same general sort as mental images: