The item difficulty and item discrimination of three multiple-choice item formats were compared in experimental and non-experimental settings. In the experimental study, 104 graduate students were randomly assigned to complete one of three forms of a multiple-choice test: (1) a complex alternative ("none of the above") as the correct answer; (2) a complex alternative as a foil; and (3) a one-correct answer format. Statistical concepts were then content tested. Analysis of variance showed significant mean differences in the item discrimination indices for the three item formats. Items having "none of the above" as the correct answer had a significantly higher mean discrimination index than did the other two item formats. There was no significant difference in the mean item difficulties for the three item formats. The non-experimental study compared the item characteristics of multiple-choice items administered as a final examination in a first-year graduate statistics course, using 44 graduate students. A significant difference in mean item difficulty was observed; the mean difference in item discrimination values was not significant. (Author/SLD)