To develop a more comprehensive picture of the variables that influence disclosure patterns, the impact of group size on sharing secrets was explored. Given Derlega and Chaikin's suggestion that the existence of a closed dyadic boundary is a prerequiste for intimate self-disclosure, it was hypothesized that subjects would be more willing to disclose intimate information in a dyad than in a triad. The results of Experiment 1, which used a role-playing methodology, confirmed the hypothesis. The main effect of group size was observed over a range of roles and items of information. In addition to the main effect, group size interaction effects also indicated that the difference between dyad and triad disclosure rates increased with more intimate items of information and with more intimate roles. These interaction effects suggested that the importance of a closed dyadic boundary depends in part on the expected confidentiality of the interchange. In Experiment 2 the conversations of groups of acquaintances were recorded and rated for intimacy. As predicted, the conversations of dyads were more intimate than those of triads. Suggestions for understanding the intimate quality of dyads are discussed.Self-disclosure is a topic that has recently generated an enormous amount of research (cf. Cozby, 1973), and substantial support for what (ourard (1971) called the "dyadic elfcct: disclosure begets disclosure" (p. 66) lias been obtained. Two hypotheses modeling and social exchange -have been offered to explain this effect. The latter hypothesis has received the lion's share of empirical support