Lindsay and Johnson (1989) and Zaragoza and Koshmider (1989) report evidence indicating that "eyewitness" subjects are much less likely to falsely claim to have seen information suggested to them verbally when they receive a source monitoring test than when they receive a recognition test requesting only identification of the seen information. The present study reports additional evidence that source misattributions are affected by the nature of the test. Intraub and Hoffman (1992) recently reported the results of a study in which subjects claimed to have seen pictures corresponding to scenes that had only been described in paragraphs they had read. With this paradigm, we found a similar effect using their test, but source confusions were reduced with a test patterned after the one used by Lindsay and Johnson. We attribute this difference in performance to the different decision criteria evoked by these two tests.