Multiple-choice testing (MCT) has several advantages which are becoming more relevant in the current financial climate. In particular, they can be machine marked. As an objective testing method it is particularly relevant to engineering and other factual courses, but MCTs are not widely used in engineering because students can benefit from guessing. In the formation of a professional engineer, guessing must be actively discouraged and this has led to this testing method being used only rarely in engineering courses. This paper describes trials of a modification to the basic MCT which is designed to assess the students' ability to use their knowledge correctly and confidently. Students are asked to select the correct answer, as usual, then to indicate their confidence in that choice. The marking scheme does not reward guessing, and misplaced confidence is penalized. Statistical results based on over 2500 answers are presented, together with survey responses. The students' attitude to the tests has been that they are fair but challenging. (Contains 1 table and 2 figures.)