A new method for quality control of multiple-choice tests was developed with the assumption that there are two classes of items-beneficial and nonbeneficial-to evaluate students' ability. The information function is applied to distinguish these two classes. Of 316 items on multiple-choice tests at Tokai Medical School in 1993, 224 (70.9%) were beneficial and 92 (19.1%) were nonbeneficial. Between these classes, the averages of passing percentage and the discrimination index were significantly different. The percentage of beneficial items varies with the form of items, whereas the average discrimination index does not.