The Naturalistic Evaluation Methodology presented here rests on a transactionist theory of child development, an ecological view of human behavior and the quasi-clinical method of inquiry. It is intended that the procedures will lead not only to the improvement of the quality of preschool education but will generate knowledge about children's behavior as a function of their interaction with the environment. Two levels of analysis have been presented that facilitate the interpretation of the data. The first, which is descriptive, involves the visual inspection of the two Behavioral Maps completed after the Behavioral Checklist has been used. (Both are included). The second, which is inferential, involves testing hypotheses concerning the effects of environmental modification on children's behavior. A time series approach based on stochastic models is advocated as the most appropriate procedure for testing these hypotheses. Procedures for data collection, determination of reliability, and data analysis are presented. Time point sampling of all children throughout the total program day is advocated, along with ratings of the environment and time sampling of the teachers' behavior. Three notions of reliability are discussed: inter-observer agreement, accuracy, and stability. (Author/CTM)