A babysitting exchange program was created for a group of women in order to build a social support network and to provide a test of the buffer hypothesis (i.e., the idea that social support may shield an individual from the negative physical and mental consequences of stress, particularly when stress is at high levels). The sample consisted of 30 inner-city women with children younger than 12 years of age. Prior to the first general co-op meeting, mothers were interviewed and pretests were administered. Measures included the Life Experience Survey, the Behavior Stress Index, and the Social Network Assessment Scale. At the first general meeting, the Symptom Checklist 90-Revised, the Tension and Depression Subscales of the Profile of Moods States, and the Proxy Measure of Health Status were administered. During the 28th week of operation, co-op members were reinterviewed and posttested. Two months after posttesting, a follow-up measure of co-op use was obtained. A steady increase in use of the co-op was observed as the women formed small groups and pairs that traded babysitting services. About half the women indicated that they found the co-op helpful. Co-op use continued after the study ended. Participation in the babysitting co-op did not have a differential impact on high- and low-stressed mothers. The implications of these findings are discussed. (RH)