BACKGROUND: In 1962 approximately 1.5 million French people living in Algeria were repatriated to France in very poor and often life-threatening conditions. These people constitute a cohort for the study of the long-term impact of gene-environment interaction on depression. AIMS: To examine the interaction between a highly stressful life event and subsequent depression, and its modulation by a length polymorphism of the serotonin transporter gene (5-HTTLPR). METHOD: A community sample of people aged 65 years and over residing in the Montpellier region of the south of France was randomly recruited from electoral rolls. Genotyping was performed on 248 repatriated persons and 632 controls. Current and lifetime major and minor depressive disorders were assessed according to DSM-IV criteria. RESULTS: A significant relationship was observed between exposure to repatriation and subsequent depression (P<0.002), but there was no significant effect of gene alone (P = 0.62). After controlling for age, gender, education, disability, recent life events and cognitive function, the gene-environment interaction (repatriation × 5-HTTLPR) was globally significant (P<0.002; OR = 3.21, 95% CI 2.48-5.12). Individuals carrying the two short (s) alleles of 5-HTTLPR were observed to be at higher risk (P<0.005; OR = 2.34, 95% CI 1.24-4.32), particularly when repatriation occurred before age 35 years (P<0.002; OR = 2.91, 95% CI 1.44-5.88), but this did not reach significance in those who were older at the time of the event (P = 0.067). CONCLUSIONS: The association between depression and war repatriation was significantly modulated by 5-HTTLPR genotype but this appeared to occur only in people who were younger at the time of exposure.