This study looks at selective memory as one of the psychological variables that could account for belief in precognitive dreaming. There is a large body of research to suggest that people tend to selectively remember information that fits with their expectations (Bartlett, 1932). The present study is an attempt to replicate Scott Madey’s (1993) findings that memory bias for ‘expectancy-confirmed’ dreams over ‘expectancy-disconfirmed’ dreams can at least partly account for the interpretation of dreams as precognitive. This investigation also extends Madey’s work by looking directly at the link between pre-existing belief in precognitive dreaming and the degree to which memory bias is exhibited. The Revised Paranormal Belief Scale was used to assess subject’s strength of belief in precognition, and then subjects were asked to read a fictional diary containing key dream-event pairs in each day’s entry. The events were either consistent or inconsistent with the dream’s predictions. Following a short distractor task subjects were asked to recall as many dream-event pairs as they could. The results confirmed expectancy-consistent information enjoys a significant recall advantage over expectancy-inconsistent information. Our analysis did not show a relationship between strength of belief and the demonstrated memory bias, although post-hoc analysis revealed those who fell in the midpoint of belief in precognitive dreaming (agnostic) had better memory for expectancy-inconsistent dream-event pairs. The results are discussed with respect to existing research in the field of memory, paranormal beliefs and dream content. We conclude by suggesting improvements to the stimulus material that might enable a more fruitful investigation into the connection between selective memory and belief in precognitive dreaming.