The typical approach to the teaching of an elementary school science methods course for undergraduate students was compared with an experimental approach based upon activities appearing in the Conceptually Oriented Program in Elementary Science (COPES) teacher's guides. The typical approach was characterized by a coverage of many topics and a combination of lectures, demonstrations, and group laboratory experiences; the experimental approach was laboratory-oriented, concentrated on investigations related to one conceptual scheme in science, and taught by involving the prospective teachers in the same type of investigations as recommended for use with the children. A semantic differential and a form of interaction analysis were used to assess, respectively, student's attitude toward science and ability in science teaching. Factor and one-way analyses were made in treatment of the pretest and posttest scores gathered in two semesters for each group for each concept, followed by an analysis of covariance on the gain scores. To assess teaching performance, the I/D and teacher talk ratios were determined by using videotape recordings of peer teaching among the subject groups. Results obtained showed that the experimental approach was superior to the typical approach and that teacher preparation should stress the teaching of investigatory type science. (Author/CC)