Female undergraduates, in groups of four, voted several times on appropriate treatment for a delinquent, using an electrical signalling device. Two simulatedgroup members consistently agreed with subjects' initial position. A third simulated member (target) exhibited one of nine response patterns. In six movement conditions (which formed a 2 × 3 design), the target (a) gradually moved a short distance toward or away from modal group opinion and (b) manifested high, medium, or low net agreement with the majority position. In three stable conditions, the target consistently (a) agreed with modal opinion, (b) disagreed, or (c) took a neutralposition. In movement conditions, the target was evaluated significantly more favourably in the toward than in the away condition and in the high agreement than in the medium and low agreement conditions. In stable conditions, the agreeing target was liked significantly better than the neutral and disagreeing targets. The target's response pattern also affected subjects' attributions about the target's motives, communication to the target (in notes interspersed between votes), and opinion change. Results were discussed in terms of previous research dealing with majority reaction to moving and stable attitudinal deviates.