Male and female observers watched videotapes of short interactions between previously unacquainted dyads of one sex or the other and then rated one target. An evaluation measure derived from the ratings was positively correlated with a majority of the diverse nonverbal behaviors coded from the videotapes. The significant correlations were all positive and all but one involved social behaviors directed toward the other person. In a multiple regression analysis, nod rate, short vocal back-channel rate, smile time, filled pause rate, and gaze time were sufficient to predict evaluations by either sex. Although no overall effects were found for sex of target, weights for behaviors varied with sex of observer in the second root of a canonical correlation analysis. Consistent with the findings of others, the amount of nonverbal activity, especially socially oriented activity, is related to favorableness of observers' evaluative impressions. The contribution of some behaviors varies with the sex of the observer.When observers form impressions of unfamiliar other persons, they have available at least four categories of immediate cues from the target: {a) verbal behavior, {b) nonverbal behavior, (c) demographic characteristics (such as race and sex), and (d) other aspects of appearance. While each type of cue has generated research interest (Secord & Backman, 1974), we find nonverbal behaviors to be especially interesting because they are both discrete public events and subtle influences on impressions. Researchers have in fact established that several nonverbal behaviors are related to impression formation. For example, Kendon (1967) and Kendon and Gook (1969) have shown that gazing is related to ratings of liking; Rosenfeld (1966) reported that approval-seeking 1. This