The purpose of this paper is to discuss some of the problems and prospects of applying production function or input-output analysis to the process of schooling. The cognitive aspect of schooling is discussed here and is restricted to those aspects readily measured by achievement tests. It is argued that education production function studies should rely less on large-scale survey data; and that they should instead develop microdata on actual processes, especially by means of the experimental identification of production functions. Applied studies in educational production functions can address two main categories of questions: (1) those dealing with efficiency, and (2) those dealing with distribution. A companion document is EA 005 152. (Author/MLF)