Presented is a study of a science program for upper elementary and middle school children encouraging the development of logical thinking skills, describing variables, interpreting and criticizing experiments, and understanding experiment design. To provide a basis for the program, the students suggested projects to be done and developed activities to correspond. Two fifth-grade classes served as the trial group and two as controls. Comparison of composite scores on the California test of Basic Skills showed comparable ability levels in the two groups. Three forms of a Science Process Test (SPT) to measure student achievement were devised. It was concluded that the personalized program was especially effective for low achievement students. Observations during the program revealed that the social structure of the classroom, and the fact that most rewards in the usual classroom came from the teacher, influenced the behavior of the students. It was also indicated that classes unaccustomed to working alone find the transition from group to personalized work difficult. (Authors/EB)