Subjects had five encounters with trained confederates, who each displayed five patterns of gaze with different subjects, in a graeco-latin square design. The patterns of gaze were: Zero, looking while talking, looking while listening, normal and continuous. Two conditions were used: Subjects were either getting acquainted with the confederates or assessing them while they behaved in an ingratiating manner. Subjects and confederates were either both male or both female. Subjects rated the confederates on fifteen rating scales; a principle components analysis produced five main components, the first two being liking/evaluation and activity/potency. Gaze affected scores on these components as predicted: Ratings of liking/evaluation increased from zero gaze to normal and were lower for continuous — confirming the affiliative balance theory; and the more gaze the higher the ratings on activity/potency. The predicted effects of ingratiation on the interpretation of gaze were not obtained, though ingratiators were seen as less intelligent. Decoders responded to amount of gaze, and not to its relation to talking and listening.