Instructors who develop classroom examinations that require students to provide a numerical response to a mathematical problem are often very concerned about the appropriateness of the multiple-choice format. The present study augments previous research relevant to this concern by comparing the difficulty and reliability of multiple-choice and completion item formats as applied to the classroom measurement of quantitative skills. This investigation also includes two variations of the multiple-choice format designed to reduce cues provided by alternatives. Focus is placed on the external validity of the experiment by using an actual examination of course material administered to students in a realistic classroom setting. When plausible distractors are used, minimal effects on difficulty and reliability are observed as a result of using "none of the above" or by using ranges of values for alternatives. The results of the study also support serious consideration of the math-completion format when efficiency of scoring is not a major concern. It is shown that fewer math-completion items are required for obtaining reliability equal to that provided by multiple-choice items. Implications which varying difficulties and reliabilities have on grading standards and test length are discussed. (Author/AL)