Fit statistics provide a direct measure of assessment accuracy by analyzing the fit of measurement models to an individual's (or group's) response pattern. Students that lose interest during the assessment, for example, will miss exercises that are within their abilities. Such students will respond correctly to some more difficult items and incorrectly to some less difficult items. Most assessment programs, including the National Assessment of Educational Progress (NAEP), currently either ignore such response anomalies or assume they do not exist. The use of a weighted-total-fit-mean-square as a measure of assessment accuracy was investigated using data from the 1990 and 1992 NAEP assessments. The distribution of fit across individuals was examined for fit and item-type differences, and the practical significance of this type of fit statistic was explored. It is concluded that this person-fit statistic has little to offer in the analysis of traditional NAEP data. Sixteen tables present analysis results. Appendix A contains 12 subscale tables, and Appendix B presents software routines. (Contains 62 references.) (SLD)