Past research has demonstrated that verbally recalling the appearance of a perpetrator after witnessing a crime can hinder one's ability to identify that perpetrator in a subsequent lineup (verbal overshadowing; Schooler & Engstler-Schooler, 1990). A recent study by Chan, Thomas, and Bulevich (2009) revealed that taking an initial memory test for an event increases one's susceptibility to later misleading information. These findings contradict those from the testing effect literature, which indicate that initial testing should enhance memory performance (Roediger & Karpicke, 2006). In the current study I investigated the effects of verbally describing a face on eyewitness suggestibility to later misinformation. Subjects witnessed a simulated crime and then either took a test over their memory for the perpetrator of the crime or performed a distractor task. Following a short delay, subjects heard misleading information about the perpetrator or only correct information. All subjects then took a final test over their memory for the perpetrator. Experiment 1 examined memory for the perpetrator using a free recall and a cued recall test. Experiments 2a and 2b examined witness identification performance with a target-present and a target-absent lineup, respectively. Three major findings emerged. First, initial testing increased correct recall probabilities and decreased misinformation recall probabilities in Experiment 1. Second, initial testing increased the likelihood of making a correct identification in the target-present lineup. Third, testing reduced identifications of individuals who matched the description of the misinformation when subjects were forced to make an identification (i.e., a biased lineup procedure). Implications for eyewitness testimony are discussed.