Over the last 15 years or so, research into measurement error has begun to consider not just whether the test items fit, but whether the people who answer the items fit. Attempts to systematically identify people who do not fit the typical pattern have led to several "person-fit statistics." This digest describes the need for such statistics, summarizes research on their use, and identifies areas in need of further research. Research has identified patterns of aberrant responses that relate to personality traits or response styles, differences in instruction, and test bias. For the most part, however, person-fit statistics have not yet been applied to many settings. Although the need has been documented, the area has been largely one of potential, rather than actual, use. While some research has addressed theoretical and methodological concerns about the application of person-fit statistics, two main questions remain: (1) whether or not they are statistically sound; and (2) whether or not they will help in practical situations. Person-fit statistics, as a logical extension of popular measurement models, are well grounded in statistical theory, but are not equally grounded with theories of learning and cognition. Research to date, however, has demonstrated that people with strange response patterns are indeed detected with few, if any, false identifications. Proponents argue that this is enough to justify routine use of this statistical tool. (SLD)