BACKGROUND: Although there is cross-sectional evidence that changes in the immune system contribute to the pathophysiology of depression, longitudinal data capable of elucidating cause and effect relationships are lacking. AIMS: We aimed to determine whether subclinical systemic inflammation, as measured by serum high-sensitivity C-reactive protein (hsCRP) concentration, is associated with an increased risk of de novo major depressive disorder. METHOD: Major depressive disorder was diagnosed using a clinical interview (SCID-I/NP). This is a retrospective cohort study; from a population-based sample of 1494 randomly selected women recruited at baseline during the period 1994-7, 822 were followed for a decade and provided measures of both exposure and outcome. Of these women, 644 (aged 20-84 years) had no prior history of depression at baseline and were eligible for analysis. RESULTS: During 5827 person-years of follow-up, 48 cases of de novo major depressive disorder were identified. The hazard ratio (HR) for depression increased by 44 for each standard deviation increase in log-transformed hsCRP (ln-hsCRP) (HR = 1.44, 95 CI 1.04-1.99), after adjusting for weight, smoking and use of non-steroidal anti-inflammatory drugs. Further adjustment for other lifestyle factors, medications and comorbidity failed to explain the observed increased risk for depression. CONCLUSIONS: Serum hsCRP is an independent risk marker for de novo major depressive disorder in women. This supports an aetiological role for inflammatory activity in the pathophysiology of depression.