This research combined experimental and correlational methods to investigate the effects of social support on social problem-solving effectiveness and perceived stress. During a wait period, college students were given the opportunity to work on practice items from a mildly stressful social problem-solving task, either alone or in the company of a close friend. Participants and friends were allowed to talk about the practice items but were not required to do so, and supportive transactions were tape recorded and content analyzed. All students then completed social problem-solving and perceived stress items alone. Participants who waited with friends did not show superior problem-solving effectiveness unless companions provided particular supportive behaviors. Participants with high perceived support scores rated the experiment as less stressful than those with low scores, but perceived support was not related to companions' actual supportive behaviors. The results were discussed in terms of implications for understanding the mechanisms of social support.