The central aim of this study was to evaluate the value of reaction time (RT) measures and event-related potentials (ERPs) for the assessment of simulated memory impairment. In two identical experiments (N = 24), healthy volunteers carried out an adapted version of the Amsterdam Short-Term Memory (ASTM) test. In half of the cases participants were instructed to perform to their best of abilities (control condition) while in the other half participants were asked to feign a believable memory deficit (simulation condition). Compared to controls, simulating participants were found to produce longer and more variable RTs. In addition, they took longer to respond to new words than to old words, while control participants showed the opposite pattern. Both early (300–500 ms) and late (500–800 ms) ERP old/new effects were similar in magnitude and scalp distribution for both groups of participants, regardless of the fact that simulating participants had significantly lower recognition accuracy than controls. The ERP old/new effect was assumed to reflect unaffected retrieval mechanisms and thus the discrepancy between this measure and recognition performance was interpreted as evidence for intentional underperformance. Results are discussed in relation to clinical observations and deception detection studies. It was concluded that the combined use of RT and ERP measures could be useful to detect and evaluate simulation of memory impairment but that future studies should include patient groups to test some of our assumptions.