The authors examined the use of mug shots as an investigative tool. Data are reported from 3 staged-crime experiments (N = 365) exploring the effects on mug-shot selections of number of faces seen before the confederate criminal's, of biased procedures, and of sorting pictures to fit the description of the criminal. The confederate was frequently selected from mug shots, and few innocent people were selected. Selections of faces declined with number of pictures viewed before the confederate's. Biased instructions and clothing bias increased choices of innocent people but not of guilty people. More innocent faces were selected when the pictures matched the confederate's description than when the pictures were not sorted to match the confederate's discription. Viewing mug shots had no effect on subsequent identifications in lineups. Implications for police use of mug shots are discussed.