A total of 21 full-day structured classroom observations were conducted in six classes of the Austin Independent School District Bilingual Preschool Project (Texas) having a high incidence of limited-English-proficient (LEP) students. Each class consisted of 15 LEP students and 3 English-proficient students. It was anticipated that English-proficient students would act as models for their LEP classmates. Two observers fluent in English and Spanish and trained in classroom observation techniques used the Early Childhood Observation Form (Revised) to record events and behaviors of specific children according to preestablished codes and definitions. Minute-by-minute records were made of the experiences of individual children across the school day. Three types of children were observed: (1) Spanish-dominant; (2) English-dominant, high proficiency; and (3) speakers of English and Spanish with low language skills. Findings indicated that over half the school day was devoted to noninstructional activities; bilingually certified teachers and their aides provided instruction; structured learning activities were conducted predominately in English, with some Spanish being used with Spanish-dominant children; teachers differed in the amount of time spent in providing structured instruction; small-group instruction was the predominant mode of instruction; and interaction between English-proficient children and LEP peers occurred during unstructured instruction. The Early Childhood Observation Form and a summary of its use in this study are appended. (RH)