This study indicates that in the context of pass/fail decisions, the use of the bock multinominal latent trait model for moderate-length tests does not produce decisions that differ substantially from those based on the raw scores. Nor does the bock model provide pass/fail decisions that are more valid than those based on raw scores when an external criterion such as teacher judgment is used. On the other hand, for very short tests the pass/fail decisions based on the bock model may differ somewhat from those decisions based on the raw scores. Thus, for very short tests, the ability tapped by the bock model appears to differ from the one implied by the raw scores. Moreover, the bock pass/fail decisions appear to relate less strongly to an outside criterion such as teacher judgment than those based on the raw scores. 1987 Helen dwight reid educational foundation.