Perceived support is consistently linked to good mental health, which is typically explained as resulting from objectively supportive actions that buffer stress. Yet this explanation has difficulty accounting for the often-observed main effects between support and mental health. Relational regulation theory (RRT) hypothesizes that main effects occur when people regulate their affect, thought, and action through ordinary yet affectively consequential conversations and shared activities, rather than through conversations about how to cope with stress. This regulation is primarily relational in that the types of people and social interactions that regulate recipients are mostly a matter of personal taste. RRT operationally defines relationships quantitatively, permitting the clean distinction between relationships and recipient personality. RRT makes a number of new predictions about social support, including new approaches to intervention. (Contains 1 figure, 1 table and 4 footnotes.)