Literature on the structure of two-person conversations has consistently found that partners become more similar i n mean duration of pauses and switching pauses over the course ot interaction. Evidence on influence in vocalization duration is primarily negative. N o direct evidence otinterspeaker influence on a moment-to-moment basis is at ailable. In this study 12 dyadic conversations are analyzed for interspeaker influence, with the use ot'time series regression procedures. Two versions ofthe data are presented: a probability suniniary including the probability of breaking mutual silences, continuing simultaneous talk, and continuing talking alone and a turn summary including vocaliration, pause, and switching pause duration. Results show that ( I ) moment-to-moment intluences are present in both versions ofthe data, ( 2 ) these intluences differ from dyad to dyad, (3) the intluences are both positive (matching) and negative (compensating), and (4) the magnitude of interspeaker influence on a temporal basis is small but detectable. In addition, there are overall tendencies to match in switching pause, probability of continued simultaneous speech, and probability of breaking mutual d e n c e s . Dyads show both compensation and matching on vocalization duration and pause-related variables. Implications of these data for past and future explanation of social interaction are explored To claim that social interaction is an interdependent process of message exchanges has become a clichCd assumption in interpersonal communication. The assumption is not without backing. Cappella ( 1 9 8 0~) reviewed over 200 studies of the influence of one speaker's expressive behavior (i.e., overt behaviors indicative of involvement) on the expressive behavior of the partner. Strong evidence of interspeaker influence across types of expressive behaviors and across developmental periods was found. Interactants' behavioral expressions of involvement seem to affect the behavioral expressions of their partners. Joseph N . Cappella (Ph.D., Michigan State University, 1974) is associate professor of communication at the University of Wisconsin, Madison, Wisconsin 53706. Sally Planalp (M.A., University of Colorado at Denver, 1977) is a doctoral candidate in communication at the University of Wisconsin, Madison, Wisconsin 53706.Despite a wealth of evidence indicating that partners affect one another's noncontent speech, affiliative behaviors, verbal intimacy, and body movements, the vast majority of studies provides little sequential information on influence between speakers. Most research has manipulated a confederate's behavior (e.g., duration of utterance) and measured the average response of subjects as a criterion. Although such designs are useful in showing the presence of influence in controlled settings, they cannot tell us whether such influence would remain in more freewheeling encounters and, by analyzing mean responses, they hide information on moment-to-moment mutual tracking and adjustment by partners.The present study focuses ...