A study examined the use of literary models in teaching written composition to children, and, as secondary goals, examined the relationships between written composition and such factors as age, sex, and intelligence. Four fifth-grade classes in Clarke County, Georgia, completed Sequential Tests of Educational Progress (STEP) Writing Tests and prepared writing samples in October, 1966, and April, 1967. For 24 weeks, the experimental group of 45 students was taught composition from models in children's literature for 1-hour periods, twice weekly. The children kept weekly records of their voluntary reading and writing. The STEP Tests revealed no significant difference in writing ability between the two groups, but the writing samples indicated some difference. Significant relationships were also found between written composition and socioeconomic status, intelligence, reading, arithmetic, and language, but none were found between composition and sex, age, or years in present school. (LH)