Prior knowledge of the likely or expected outcome of a forensic investigation has been shown to produce biases in the results obtained, reducing objectivity. The wide prevalence of such cognitive biases in many judgments has long been recognised by social psychologists, but its importance is only now gaining appreciation within forensic science communities. It is therefore timely to draw attention to the power of cognitive biases found in a study of the influence of administrator expectations on photographic identifications. Data are presented to show that when a line-up administrator knows the identity and position of a target within a line-up choice, in which the 'witness' is ignorant of the actual target, that target is more than twice as likely to be selected compared with when the administrator is kept 'blind'. These findings, taken together with related studies, support the recommendation that all forensic analyses are made 'double-blind'-a method that has proven to be effective in reducing such effects within the social sciences.