As an attempt to confirm the view that the relationships between certain classroom process variables and student achievement are curvilinear, first grade reading and mathematics data from the Stallings and Kaskowitz (1974) Project Follow Through Classroom Observation Evaluation were reanalyzed using orthogonal polynomial regression methods. Seventy-five classroom process variables were selected from Stallings and Kaskowitz data and entered into poly-nomial regression analyses. Thirty-two of ninety-five relationships tested were found to be nonlinear. (This, of course, is well beyond chance.) Fourteen of the nonlinear functions were quadratic in nature, eleven were cubic, and seven were quartic. Fifteen of these functions were related to mathematics achievement, and seventeen were related to reading achievement. These results generally support the view that relationships between some classroom process variables and student achievement are probably curvilinear. In turn, this finding seems to raise a question about recommendations for instructional improvement that are made based solely on linear analyses of process-product study data. This finding also provides additional support for the argument to employ both linear and nonlinear analyses in classroom process-product research. (Author/MP)