Previous research on the buffering effects of social support focused mainly on life events as stressors, and mental illness as outcome. Furthermore, the question as to why support influences illness has not been subjected to theoretical or empirical study much. In this article we develop a hypothesis on the basis of the theory of social capital. We hypothesize that specific types of social resources are more relevant to the consequences of some events than of others. We test this hypothesis in two ways: (1) by taking life events as stressor and occurrence of illness as outcome, and, which is somewhat unusual, (2) by taking illness as stressor and duration and disabilities of illness as the outcome. Analyses of a representative sample of the Dutch population (N = 10,110) reveal that receiving specific types of support does not lead to better health or less illness in cases of stress. On the contrary, people who are under stress and receive more support, also appear to report more illness, more disabilities and a longer duration. We suggest that in an open sample like ours, the disease level measured is not severe enough to assess buffer effects of social support.