College students were given the Multidimensional Self-Esteem Inventory. For each subscale of the inventory, they were classified as having either high self-esteem (top 40% of the scorers) or low self-esteem (bottom 40% of the scorers). Several weeks after the assessment of self-esteem, they watched a 60-s videotape in which a model maintained eye contact with an interviewer for either 5 s or 50 s. After viewing the tape, the students again completed the Multidimensional Self-Esteem Inventory, but this time as they thought the model in the tape would complete it. The general pattern of the results was that high self-esteem students perceived greater self-esteem in the model in the 50-s tape compared to the 5-s tape, whereas low self-esteem students perceived greater selfesteem in the model in the 5-s tape compared to the 50-s tape. The results show the importance of an observer's own self-esteem in the formation of impressions based on eye contact. Implications for further research are discussed.Maintaining eye contact in situations such as a job interview is a characteristic that generally creates a favorable impression for an observer (e.g., Wheeler, Baron, Michell, & Ginsburg, 1979; Wiens, Harper, & Matarazzo, 1980). The early studies investigating eye contact as a social variable were correlational; participants view videotapes, photographs, or actual interview situations and then make ratings of the characteristics of the person. These ratings are then correlated with patterns of eye contact.More recently, Brooks, Church, and Fraser (1986), Brooks (1993), andNapieralski, Brooks, andDroney (1995) conducted experimental studies manipulating duration of eye contact in a social situation. Specifically, 60-s tapes were made in which a model was listening to instructions from another person. Three separate tapes were made, with the model maintaining eye contact for either 5 s, 30 s, or 50 s. The tapes were shown to college students who then gave personality assessments of the individual in the tape. As eye contact increased, the model in the tape Requests for reprints should be sent to Charles I. Brooks,