Although social support (SS) is associated with risk for major depression (MD), we are uncertain of the extent to which a) low SS increases risk for MD, b) MD lowers SS, or c) both variables reflect a common genetic liability. Using two waves of interview data on female twin pairs from a population-based registry, we examine the cross-sectional and longitudinal association of eight dimensions of perceived SS and MD. Risk for MD in the last year was inversely associated with supportive spouse and relative relationships, and directly associated with problems in these relationships (e.g., too many demands, criticism, tension, and disagreements). Significant cross-time associations were seen only for spousal variables. The history of MD in one twin significantly predicted low relative and spouse support, and relative and friend problems, in her co-twin (MZ > DZ pairs). The relationship between SS and MD in women is complex and due to at least the three separate mechanisms outlined above that operate to varying degrees in different dimensions of SS.