Context: Epidemiological findings demonstrating an increased risk for individuals with social anxiety disorder (SAD) to develop depression have been challenged by discrepant findings from prospective longitudinal examinations in childhood and early adolescence.

Objectives: To examine patterns of SAD incidence, the consistency of associations of SAD with subsequent depression, and distal and proximal predictors for subsequent depression.

Design: Face-to-face, 10-year prospective longitudinal and family study of up to 4 waves. The DSM-IV Munich-Composite International Diagnostic Interview was administered by clinically trained interviewers.

Setting: Community sample in Munich.

Participants: Three thousand twenty-one individuals aged 14 to 24 years at baseline and 21 to 34 years at follow-up.

Main outcome measures: Cumulative incidence of SAD and depression (major depressive episode or dysthymia).

Results: Cumulative incidence for SAD was 11.0%; for depression, 27.0%. Standardized person-years of incidence for SAD were highest for those aged 10 to 19 years (0.72%) and were low before (0.20%) and after (0.19%) that age range. Depression incidence was different, characterized by delayed and continued high rates. Social anxiety disorder was consistently associated with subsequent depression, independent of age at onset for SAD (relative risk range, 1.49-1.85, controlling for age and sex). Crude Cox regressions showed significant distal (eg, parental anxiety or depression, behavioral inhibition) and proximal SAD characteristics (eg, severity measures, persistence) as predictors. Most associations were attenuated in multiple models, leaving behavioral inhibition (hazard ratio, 1.30 [95% confidence interval, 1.04-1.62; P = .02]) and, less consistently, panic (hazard ratio, 1.85 [95% confidence interval, 1.08-3.18; P = .03]) as the remaining significant predictors.

Conclusions: Social anxiety disorder is an early, adolescent-onset disorder related to a substantially and consistently increased risk for subsequent depression. The demonstration of proximal and particularly distal predictors for increased depression risks requires further exploration to identify their moderator or mediator role. Along with previous evidence that comorbid SAD is associated with a more malignant course and character of depression, these results call for targeted prevention with the aim of reducing the burden of SAD and its consequences.