Concern with the development of formal reasoning has been stimulated by the increasing recognition that formal reasoning is related to student success in secondary and college science and mathematics courses (Arons, 1976;Bauman, 1976;Griffiths, 1976;Herron, 1975;Karplus, 1978;Kolodyi, 1975;Lawson and Renner, 1975;Sayre and Ball, 1975;Shayer, 1972Shayer, , 1973Shayer, , 1974Suarez, 1977). In this article we shall report on our study of six group-administered tasks that have not been previously used in the United States with student groups of widely differing ages. We seek to answer these questions:1. What categories are required for classifying the subjects' responses on tasks requiring proportional, probabilistic, or correlational reasoning?2. How effective are these tasks for assessing these aspects of formal reasoning? 3. What implications for teaching are suggested by the observed distributions of student responses among the categories required? SUBJECTS The subjects were 505 students from sixth grade to college sophomores, ranging in age from 11.5 to 20.0 years. Somewhat over 400 of the subjects were enrolled in elementary and secondary schools located in a middle-to upper-middle-class suburban community in the San Francisco Bay area. The eighth graders were tested in English classes, the tenth graders in biology, and the twelfth graders in social studies to obtain representative samples not biased with respect to mathematics and science background. The college students were freshmen and sophomores enrolled in two physical science courses for non-majors at the University of California, Berkeley. The numbers and grade levels of the students are summarized in Table 1.