The fallibility of eyewitness identifications is well documented. Nevertheless, research has yet to assess the possibility that the type of crime committed systematically influences who eyewitnesses mistakenly identify. We address this oversight by presenting a contextual model of eyewitness identification (CMEI). The CMEI asserts that discrete crimes automatically activate distinct stereotypes about a perpetrator's appearance. Depending on the congruence between these stereotypes and the perpetrator's actual appearance, eyewitnesses will remember the perpetrator as appearing more (or less) representative of his or her group (i.e., higher or lower on perceived stereotypicality). Estimator and system variables are posited to affect identifications at different stages of the identification process. The literatures on stereotype activation, perceived stereotypicality, and stereotype-consistent memory biases are reviewed to support the CMEI. Our conceptual integration provides a model of eyewitness identification that explains when mistaken identifications are likely to occur and who they are likely to affect. ). Although this literature has done an excellent job of documenting when errors in eyewitness identification are likely to occur, little is known about whether the type of crime committed (i.e., the crime type) systematically affects who eyewitnesses mistakenly identify. In the current paper, we argue that focusing on the crime type can help explain why eyewitnesses choose one innocent suspect over another from a police line-up. To advance this thesis, we provide an integration of the literatures on (i) the automaticity of stereotype activation, (ii) perceived stereotypicality, and (iii) stereotypeconsistent memory biases. In doing so, we develop a contextual model of eyewitness identification (CMEI). After showing how these respective literatures support the CMEI, we review recent studies that directly test our model. We conclude by discussing the generalizability of the CMEI and examining the implications our synthesis has for the identification process.