Errorless learning, i.e. learning under conditions that prevent the generation of false memory candidates, has been shown to lead to enhanced memory performance during retrieval compared to errorful, trial-and-error, learning. These two learning methods were implemented in a word-stem completion format and contrasted in young healthy subjects with stimulus-locked and response-locked event-related brain potentials (ERPs) being recorded during memory retrieval. Retrieval accuracy was enhanced for items studied under errorless conditions. Stimulus-locked ERPs showed a modulation of a right frontal effect that differed in amplitude and topography for items learned under errorful and errorless conditions. This effect is interpreted as reflecting post-retrieval processes. In the response-locked ERPs, a typical error-related negativity (ERN) was observed that was most prominent for false alarm trials in the errorless condition, of medium amplitude for hits and false alarms in the errorful condition, smaller for hits in theerrorless condition and virtually absent for items correctly or incorrectly classified as new. We propose that the amplitude modulation of the ERN is related to the activity of an internal monitoring device checking the veridicality of a retrieved memory trace.