Conflicting models persist over the nature of long-term memory. Crucial issues are whether episodic memory and recognition memory reflect the same underlying processes, and the extent to which various brain structures work as a single unit to support these processes. New findings that have resulted from improved resolution of functional brain imaging, together with recent studies of amnesia and developments in animal testing, reinforce the view that recognition memory comprises at least two independent processes: one recollective and the other using familiarity detection. Only recollective recognition appears to depend on episodic memory. Attempts to map brain areas supporting these two putative components of recognition memory indicate that they depend on separate, but interlinked, structures.