This study examined the role of social supports and life stress events in use of mental health services. Data were derived from a 3-5 year longitudinal study of a general population of adults (ages 14 and older) enrolled in the RAND Health Insurance Experiment. The sample used in this analyses included those enrollees who participated for a full year in the second (N = 1979) and third (N = 2601) years of the experiment. Included were 4580 person-years of data. Statistical methods used probit regression models which controlled for a number of covariates including insurance plan, site, sociodemographic variables, physical and mental health. Results supported the following conclusions: (1) the more life events experienced, the more likely one is to use mental health services, (2) chronic types of life events are more important than acute events in predicting use of mental health services, (3) when defined as social contacts, social support does not predict use; however, when defined as social resources, the more support one has the less likely is one to use mental health services, and (4) neither social contacts nor social resources buffer the impact of life stress events on use of services.