How the brain supports the storage of information into long-term, episodic memory is an important, yet unresolved, question. This presentation considers findings from recent ERP and fMRI experiments in which episodic encoding is studied via the ‘subsequent memory’ approach. In this approach, neural activity evoked by information presented to subjects in a study period is classified according to whether the information is remembered or forgotten in a subsequent memory test. Such ‘subsequent memory’ effects differ qualitatively depending on the type of task subjects perform at study, and the way in which memory is probed. Subsequent memory effects can be seen for transient neural activity associated with processing individual items as well as for neural activity that is sustained throughout a task. It seems that the brain does not support episodic encoding through the same neural system under all circumstances. Rather, there appear to be multiple encoding systems, the employment of which depends on the nature of the processes engaged by information at initial exposure, those engaged during subsequent retrieval, and the degree of overlap between the two. Moreover, effective encoding seems to rely not only on how individual items are processed, but also on the mental state during which those items are experienced.