The Maryland Functional Testing Program (MFTP) uses the Rasch model as the statistical framework for the analysis of test items and scores. This paper is designed to assist the reader in developing an understanding of the fit statistics in the Rasch model. Background materials on application of the Rasch model in statistical analysis of the MFTP are provided, studies of item fit and causes of item misfit are outlined, and recommendations for uses of fit statistics are made. It is recommended that: (1) since the number of items detected as "misfitting" is a matter of choice in selecting the fit statistic from a wide range of fit statistics produced by BICAL, and since there are no sound statistical procedures for selecting items on the basis of fit indices; test of item fit should not be used with some arbitrary critical value to make automatic decision for eliminating them; (2) items that have total or between "t" values larger than 2.00 should be investigated for flaws in the construction of the item, unusual item content, "order effects" resulting from its position in the test, and level of difficulty; (3) if an easy item does not fit the model because it occurs near the end of the test which low ability students may not reach time for the test might be extended; (4) guessing is inherent in the use of multiple-choice items and such items should not be deleted for this reason; and (5) items flagged by various tests of fit should be studied by subject matter experts to generate possible hypotheses about the reason for misfitting items. Fit statistics produced by BICAL-1980 are described, with emphasis on the sample size effect on fit indices, and computation of these statistics are explained. One data table is included. (TJH)