An attempt is made to explore the use of subjective probabilities in the analysis of item data, especially criterion-referenced item data. Two assumptions are implicit: (1) one wants to obtain a maximum amount of information with respect to an item using a minimum number of subjects; and (2) once the item is validated, it may well be administered in the classical correct/incorrect manner. One way to satisfy these assumptions is to initially administer the unvalidated item to a small number of subjects using confidence testing procedures. Then subjective probabilities can be translated to pseudo-classical item scores which are, at least in theory, guessing-free. Using pseudo-classical scores, a relatively sophisticated item analysis table can be constructed, typical item statistics can be calculated, and a kind of item reliability independent of total test reliability can be assessed. As a useful bridge between subjective probabilities and classical correct/incorrect scores, pseudo-classical scores appear to be of potential use in the analysis of criterion-referenced items. (Author/RC)