This paper discusses the use of process-oriented research in early childhood programs. Process analysis is contrasted with outcome-oriented evaluation, the latter seen as prevalent in current research practice. Four functions of process analysis are: to describe current teaching practices, to train teachers, to monitor instructional programs, and to investigate relationships between classroom activities and pupil growth. Systems for observing the process of teacher-child interaction are categorized as either open, low-inference systems or closed, high-inference systems. The specimen-description technique is cited as an example of the forner, while examples of closed systems are time-sampling, event-sampling and trait-rating. Methodological issues discussed relate to choice of appropriate instruments, reliability of observers, and degree of researcher control over selection and supervision of the program to be evaluated. (Author/BF)