Many studies show that witnesses can develop false memories for suggested misinformation provided by an interviewer. The forced fabrication effect extends this finding by demonstrating that witnesses can also develop false memories for events they were forced to fabricate themselves. In two experiments we compared the incidence of false memory following forced fabrication and interviewer provided suggestion under various conditions (pre-test warning/no warning; one-week/two-week delay) and type of test (source recognition vs. narrative recall). Whereas interviewer suggestions resulted in more false memories than forced fabrications on source recognition tests when participants overtly resisted fabricating and were warned at test, tests of narrative recall showed the opposite pattern even with pre-test warnings: Fabrications generated following overt resistance led to more false recall than interviewer provided suggestions. This dissociation between suggestive interview type and test type indicates that predictions about the deleterious consequences of interviews are dependent on the way memory is assessed. Copyright © 2011 John Wiley & Sons, Ltd.