The purpose of this study is to examine some fundamental dimensions of expression of nonverbal behavior (NB) through analysis of dyadic interactions. Twenty pairs of subjects were divided into four experimental groups; (male pair, female pair) × (unfamiliar, familiar). They were instructed to talk with each other about “a part time job” for fifteen minutes, and this was recorded by two video cameras. Two minute interactions were sampled from both pre-instruction situation and fifteen minute conversation. The frequencies, total durations, and mean durations of 8 NB items (gaze, smile, leaning forward, leaning backward, turning away, speech, silence, nodding by speaker) and 8 unit NB (UNB) items (nodding by listenter, smile back, reaction to speech, silent synchrony, imitative synchrony, following synchrony, concurrent synchrony, mirroring posture) were measured. As for“nodding by speaker” and UNB items apart from “posture mirroring”, only frequencies were measured. Factor analysis of all behavioral variables yielded four factors; Contact, Relaxation, Approach, Avoidance. Sequential analysis which describes succeeding behavior patterns indicated that there are individual specific pattcrns; from Silence (that is, not Contact) to Contact, from Contact to Relaxation, from Avoidance to Approach. As a result, I assumed three dimensions in the expression of NB; “Contact-Silence”, “Relaxation-Tension”, and “Approach-Avoidance”. Correlation between scores on each dimension and each psychological index (e. g. estimation of affects, estimation of the other person) characterized each dimension. Multiple regression analysis indicated that the three dimensions accounted for 73.4% of “degree of involvement” which is related to important NB function. Therefore the validity of explaining expressive aspect of NB in terms of these three dimensions was supported. The result from analysis of only NB items (except UNB items) was almost the same as that from analysis of all items. As sequential analysis shows, the same behavior is often repeated between individuals, which suggested that most NB often constitutes UNB. That seemed to be why UNB didn't make any distinctive contribution.