Social support is consistently associated with health and adaptation, yet the concept is often treated too simply. Network availability, perceived support, received assistance, companionship, informational exchange, practical help, relational responsiveness, and institutional navigation are related but non-equivalent phenomena. This narrative review re-examines social support across health and illness by integrating evidence on construct differentiation, psychological mechanisms, clinical applications, relational costs, cultural context, and inequity. The central argument is that support should not be understood as a uniformly beneficial resource whose effects increase with quantity. Its relevance depends on what support is available, what is actually exchanged, how the recipient interprets it, whether it fits the stressor or illness phase, the obligations or constraints accompanying it, and the structural conditions governing access and use. Evidence for stress buffering is also outcome-dependent: supportive relationships may relate to emotional regulation, coping, self-efficacy, health behavior, and selected biological processes, but null and heterogeneous findings challenge universal mechanistic claims. The review therefore develops a bounded, multilevel account in which structural opportunity, network configuration, relational exchange, subjective appraisal, psychological processes, behavior, and healthcare participation remain analytically distinct. Particular attention is given to perceived–received support mismatches and to unequal capacities to obtain useful support. Clinical translation should consequently prioritize fit, responsiveness, accessibility, and context rather than simply increasing contact or assistance.
