Reported are background information and initial analyses of Tennessee's Project STAR, a legislatively authorized study of class size. The aim of the project was to identify causes and effects of differences in elementary school class size. Primary analyses were conducted on kindergarten data for 1985-1986. During that year, 42 school districts and 79 individual schools participated in the project. Core variables were three classroom conditions: (1) small class, defined as one teacher with about 15 students; (2) regular class, defined as one teacher with about 26 students; and (3) regular class with a teacher aide. These three classroom conditions were investigated: (1) in inner-city, suburban, urban or small city, and rural locations; (2) in relation to pupil variables (race, sex, socioeconomic status); and (3) in relation to either achievement measures (mathematics, sounds and letters, words and sentences, total reading scores) or noncognitive measures (motivation, self-concept). While the unit of data collection was the individual pupil, the primary unit of analysis was the class. The report presents tables and information concerning two comparisons among class types and three comparisons among the four locations. Findings seemed to favor the small class condition slightly. Supplementary analyses are briefly reported. (RH)