The New York State Department of Civil Service investigated an empirical application of the Rasch model which appears useful in Civil Service testing. The model is a powerful tool for developing insights into what test items are measuring while permitting the investigator to spot defective items. It also reveals meaningful distinctions in the type of task set by different items. Two sets of specific examples are discussed to illustrate its usefulness. The first set of examples considers the use of item probability as an index of the degree of fit of the material to the model, while the second discusses the "normal deviate" matrix, which displays the goodness of fit of each item at each score group, and enables an investigator to ascertain the overall validity of the general index. These examples demonstrate the applicability of the Rasch model to a variety of conditions. The author suggests that the model seems promising for civil service testing since it is not simply a means to derive scores but is also a powerful tool for test analysis, construction, and design. (CK)