Thirty four-year-olds and 30 five-year-olds participated in a study of social influences on computer activity. During a first session children were individually pretested for general cognitive level and alphabet and number skills; during a second session they were given individual training on the computer; during third and fourth sessions half of the childen were observed while working alone at a computer and half were observed while working with a peer at a computer; and during a fifth session each child was individually tested for retention and feelings about the computer experience. In general, behavior of paired children was more engaged and more effective than that of nonpaired children, but was not more efficient. In addition, children who worked with a peer were observed showing more positive affect, rated their affect higher, and retained more about the experience than children who worked alone. Research conclusions also suggested that the advantages of peer interaction increased over the preschool years and that dyadic computer activity is educationally feasible and useful for young children. The results were discussed in terms of a four level-framework of social influence which hypothesized that subjects' skill level relative to task difficulty determines the nature of social influence. (DST)