System-variable research investigates variables that are manipulable in actual criminal cases (e.g., the structure of a lineup) and thus has the potential for reducing the inaccuracies of eyewitnesses. Estimator-variable research, however, investigates variables that cannot be controlled in actual criminal cases (e.g., characteristics of the witness) and thus can only be used in the courtroom to augment or discount the credibility of eyewitnesses. When system variables and estimator variables are contrasted with respect to their relative potential for positive contribution to criminal justice, system-variable research appears to have greater applied utility for criminal justice than in eyewitness identification. Nevertheless, estimator-variable research may circumvent application problems by using the individual as a unit of analysis rather than focusing on the situation as the unit of postindictive analysis. Specifically, self-report information may be diagnostic of accuracy. This cognitive approach assumes that situational factors prior to and during the crime affect some measurable response that is statistically related to the accuracy factor. Seven notes and 33 references are supplied. (Author abstract modified)