Multidimensional item difficulty (MID) is proposed as a means of describing test items which measure more than one ability. With mathematical story problems, for instance, both mathematical and verbal skills are required to obtain a correct answer. The proposed measure of MID is based upon three general assumptions: (1) the probability of answering an item correctly increases monotonically with an increase in each dimension being measured; (2) it is desirable to locate an item at a single point in a multidimensional space; and (3) the best point to use in defining the MID for an item in the multidimensional space is the point where the item is most discriminating and provides the most information about the person being measured. To demonstrate the procedure, the multidimensional extension of the two-parameter logistic (M2PL) model has been selected. M2PL parameter estimates are computed using item response data from a sample of 1,000 students who took the ACT Assessment Mathematics Usage Test. As a basis for comparison, a traditional item analysis and a three-parameter logistic calibration using LOGIST are performed on the same data set. It is concluded that the results are readily interpreted. (GDC)