The proposal that the hippocampus is important for the encoding of episodic information, but not familiarity-based recognition, is incompatible with the available data. An alternative way to think about functional specialization within the medial temporal lobe memory system is suggested, based on neuroanatomy.