Using Guttman's facet design analysis, four parallel forms of a multiple-choice test were developed. A mapping sentence, logically representing the universe of content of a basic cardiology course, specified the facets of the course and the semantic structural units linking them. The facets were: cognitive processes, disease priority, specific disease, and clinical judgment stage. Developing items from specification strings was simpler and quicker than developing items from objectives without specific guidelines. Due to logistical problems, only two of the four parallel forms originally planned were actually used. Although this increased the number of times each item was used, it did not violate the theoretical methodology, nor the notion that the tests were parallel. The tests will be field tested with fourth-year medical students at five medical schools, to establish their parallel forms reliability. (CP)