The three variables of interest were the opportunity to view, description accuracy, and witness confidence. The subjects for the study were forty-seven female bank tellers who were chosen at random. The Florida attorney general's office developed a suspect identity chart which was used in the study. A target person entered each bank and tried to cash an altered money order. Several hours after the incident each teller was interviewed. The results showed that description congruence, description accuracy, and description completeness are not predictive of identification accuracy in a meaningful field setting. It seems that witnesses' confidence in their positive identifications may carry more information about the likely accuracy of those identifications than do witnesses' confidence in their decision that the target was not in the lineup. This study provides evidence of the independence of identification accuracy and description quality in a forensically relevant field setting. This finding is directly applicable to the court system, some of which have stressed that social science research carried out in a forensically meaningful situation can be of considerable value to the courts' decision-making duties. 22 references, 1 table.