In the first context, 100 female students of health-related professions were given a description and asked to choose the most likely suspect and to estimate certain gross physical features of the persons depicted in the photospread. In the second context, 65 female undergraduates from an introductory psychology course served as mock witnesses and were asked only to estimate the gross physical features of the persons, thus removing the influence that a description and choice might have on feature estimates. Context was a critical consideration, as mock witnesses appeared to be influenced by the description given and their task of reporting weight estimates. The defendant was chosen by 58 percent in one photospread rated as biased by researchers because it did not contain enough plausible foils. The article recommends that eyewitnesses trying to make an identification from a line-up be given the option of not choosing any line-up member. Tables, photographs, and 13 references. (Author abstract modified)