A study examined the role of prior knowledge in teacher-generated analogies and student-generated analogies that convey the concept of protein synthesis. Subjects were 269 eighth-grade students enrolled in a public school district in a northern suburb of Chicago, Illinois. Subjects were divided into groups that received traditional instruction, generated their own analogies, or received teacher-generated analogies for instruction. Student performance was measured using a publisher's multiple-choice test and a short-answer test to assess knowledge about protein synthesis before instruction, immediately after instruction, and one month after instruction. Results indicated that in both the immediate posttest and the one-month posttest, prior knowledge was a significant variable in predicting the student's performance on the posttests. Findings suggest that a foundation of science knowledge seems requisite to the efficient and effective use of teacher-generated analogies as well as student-generated analogies. Teachers should take pains to use analogies based on domains already familiar to the students and should make clear the semantic and structural correspondences between the analogs that are important for providing the desired explanation. (Contains 16 references and 2 tables of data.) (RS)