Historically, the hippocampus has been viewed as a temporary memory structure. Consistent with the central premise of standard consolidation theory (SCT), a memory is initially hippocampus-dependent but, over time, it undergoes a consolidation process and eventually becoming represented in a distributed cortical network independent of the hippocampus. In this paper, we review evidence that is incompatible with each of the following essential features of SCT that are derived from its central premise: (1) Hippocampal damage reliably produces temporally graded retrograde amnesia, (2) all declarative explicit memories are equivalent with respect to consolidation, (3) consolidation entails a process of duplication in which a particular cortically based memory is identical to the hippocampus-dependent memory from which it derived, (4) consolidated memories are permanent and immutable. We propose an alternative hypothesis that assumes a transformation process and changes in the memory over time. Building on multiple trace theory (Nadel & Moscovitch, 1997), the transformation hypothesis contains three key elements that differentiate it from SCT: (1) An initially formed memory, which is assumed to be episodic and context-bound, remains dependent on the hippocampus for as long as it is available, (2) with time and experience, a hippocampal memory supports the development, in neocortex, of a less integrated, schematic version, which retains the gist of the original memory, but few of its contextual details, (3) there is a dynamic interplay between the two types of memory such that one or another may be dominant, depending on the circumstances at retrieval. Evidence is provided in support of the transformation hypothesis, which is advanced as a framework for unifying the seemingly disparate results of studies of anterograde and retrograde memory in the animal and human literatures.