Task manipulations and a subtractive analytic strategy were combined to decompose the positive-going amplitude shift of the event-related potential (ERP) that is elicited by previously studied words during recognition memory testing. Target stimuli elicited a robust posterior P3b subcomponent, but this effect is shown to be independent of the effects of prior study. Two study-related subcomponents were also identified. The earlier study effect reached peak amplitude about 400 msec after word onset and was elicited by repeated words regardless of whether the task required retrieval of information from secondary memory. This effect might reflect a repetition-related decline in the amplitude of the N400 component. The other study-related effect reached peak amplitude around 600 msec post-stimulus onset. This later effect overlapped temporally with the P3b, but it was not related to targetness. Rather, it was specifically related to task conditions that required retrieval from secondary memory. The identification of this third effect helps to clarify the functional significance of the ERPs observed during recognition memory testing.