The purpose of this study was to evaluate modes of art instruction for children in kindergarten. These modes were derived from an experimental art instruction program. The experimental modes consisted of all possible combinations of the following instructional components: (1) a series of charts demonstrating Line, Shape, Color, and Texture-Pattern; (2) art consultants; and (3) special in-service training for teachers. At the extremes, the full program as one mode was compared to a spontaneous, unstructured art learning situation as another mode. The sample consisted of 30 kindergarten classes in the Grand Rapids Public School System. They were selected from high, medium, and low socioeconomic school districts. The data were analyzed by means of complex analysis of variance with mode of instruction, socioeconomic status, and length of instruction as the major variables (main effects). Data (art productions) were evaluated by trained art personnel by means of standardized criteria. Statistical reliability among evaluators was computed and found to be high. This study found that mode of instruction and SES were significant. The interaction between SES and mode was also significant. The experimental program had less effect for low SES than high SES children. The length of instruction was not significant. The full experimental mode of instruction was most successful in developing superior art productions. (Author/DB)