My central argument is that the theoretical presuppositions of psychology serve to obscure the debate over false memory. Psychologists have failed to recognize that the reliability of personal memory is a presupposition human knowledge; we have to ado pt a non-neutral stance to personal memory-claims as testimony. Memory may not be "reproductive," but it is not "reconstructive" either, in the sense defended for instance by Elizabeth Loftus in The Myth of Repressed Memories. Philosophical account s of personal memory emphasize the distinction between personal and merely factual memory of one's own past, which tends to be neglected in psychology. This is a datum, not a result of theorizing. I elucidate the distinction, and contrast di rect and indirect realist theories of memory, defending the former as emphasizing the spontaneous, authoritative nature of personal memory-claims. The "reconstructive" conception of "memory as narrative" assumed by psychologists such as Loftus disparages reliability and spontaneity. It claims that I integrate my personal memories with less direct sources of knowledge into a "narrative" that appears consistent to myself now. This view has not been much discussed in the philosophical literature. I argue tha t the "memory as narrative" conception is mistaken, and serves to confuse the debate over false memory. I also argue that the debate is unlikely to be undermined by Ian Hacking's argument that past human actions are to a certain extent indeterminate. My p oints of disagreement with Loftus concern: (1) Her mis-assimilation of false memory with ordinary memory-errors. Where memories are mistaken, they are almost invariably "false-in-detail". "Completely-false" memories are necessarily rare. Loftus and other psychologists tend to gloss over this vital distinction. The results of therapeutic suggestion are not part of a general pattern of "reconstruction" of the past through personal memory, as the narrative conception suggests. (2) Her wholesale rejection of repression. However, she may be right in arguing that there is a distinct and questionable category of "robust repression." (3) Her conviction that scientific theory can make us change our ordinary methods of assessing the reliability of tes timony. There are strict limits to such change. I conclude by examining the tension between repression and the paradigm of a spontaneous memory-judgment.