Recognition of faces from still photographs was measured asa function of whether the faces were presented in positive or in negative during the initial viewing and subsequent recognition procedures. Recognition accuracy was significantly lower when faces were initially viewed in negative, regardless of their mode of presentation during the recognition procedure. The results provide further support for the hypothesis that still photographs of faces are recognized on the basis of properties specific to faces, rather than simply as patterns, and suggest that the information stored during the viewing of faces in negative cannot readily be transformed into its faces-in-positive corollary.