Choices from arrays are often characterized by position effects, such as edge-aversion. We investigated position effects when participants attempted to pick a suspect from an array similar to a police photo lineup. A reanalysis of data from 2 large-scale field studies showed that choices made under realistic conditions—closely matching eyewitness identification decisions in police investigations—displayed edge-aversion and bias to choose from the top row (Study 1). In a series of experiments (Studies 2a–2c and 3), participants guessing the location of a suspect exhibited edge-aversion regardless of whether the lineup was constructed to maximize the chances of the suspect being picked, to ensure the suspect did not stand out, or randomly. Participants favored top locations only when the lineup was constructed to maximize the chances of the suspect being picked. In Studies 4 and 5, position effects disappeared when (a) response options were presented in an array with no obvious center, edges, or corners, and (b) instructions stated that the suspect was placed randomly. These findings show that position effects are influenced by a combination of task instructions and array shape. Randomizing the location of the suspect and modifying the shape of the lineup array may reduce misidentification.