Recent studies have shown that involuntary autobiographical memories (IAMs) can be elicited in the laboratory. Here we assessed whether the specific instructions given to participants can change the nature of the IAMs reported, in terms of both their frequency and their characteristics. People were either made or not made aware that the aim of the study was to examine IAMs. They reported mental contents either whenever they became aware of them or following a predetermined schedule. Both making people aware of the aim of the study and following a fixed schedule of interruptions increased significantly the number of IAMs reported. When aware of the aim of the study, participants reported more specific memories that had been retrieved and rehearsed more often in the past. These findings demonstrate that the number and characteristics of memories depend on the procedure used. Explanations of these effects and their implications for research on IAMs are discussed.