Event-related brain potentials (ERPs) and performance measures were recorded from young adults during continuous recognition memory for word or picture representations of the same concepts. Subjects made speeded choice responses as to whether the item was "new" or "old," with "old" items repeated after lags of 2,8, or 32 intervening stimuli following their first presentation. Longer lags were associated with poorer performance in words but not in pictures. A sequence of endogenous ERP components was responsive to repetition (old vs. new) and stimulus modality (pictures vs. words), and differed in scalp distribution as a function of both variables. The data suggest that multiple brain generators contribute to ERP repetition effects and support theoretical models that hypothesize processing differences between words and pictures.In conjunction with behavioral measures, the recording of event-related brain potentials (ERPs) has been shown to be a useful tool for mental chronometry, enabling more direct mapping of theoretical concepts derived from cognitive psychology onto nervous system activity. We here report on the behavioral and ERP correlates of continuous recognition memory, focusing on the effects of stimulus modality (word vs. picture presentation of a concept), item repetition (old vs. new), and lag (i.e., the number of intervening items between first and second presentation).It is currently unclear to what extent different brain systems are involved in the processing of words and pictures. The empirical finding that memory for pictures is usually superior to memory for words has been hypothesized to result from a more distinctive sensory code for pictures (Nelson, 1979) or because pictures are more likely than words to be encoded by dual verbal and nonverbal coding systems (Paivio, 1986). To the extent that word and picture memory show similar correlations with experimental variables, support can be claimed for theoretical positions that postulate amodal cognition or minimal differences between the processes used to encode and retrieve word and picture stimuli. Since differences in scalp distribution between conditions are usually cited as evidence in support of differences in the locus of intracranial generation, finding such word-picture differences might be indicative of more separable brain and information-processing systems for words and pictures.Repetition effects may represent operations primarily associated with the first presentation, such as initial schema activation, which need not be as detailed on second presentation, or retrieval-related processes, such as