OBJECTIVE: To analyse in the general population the prevalence and predictors of symptomatology consistent with post-traumatic stress disorder (S-PTSD) 18 months after an industrial explosion. DESIGN: Cross-sectional survey. PARTICIPANTS AND OUTCOME MEASURES: A random sample of 1191 city inhabitants, including an oversample of the immediate area (<3 km). S-PTSD was measured by the self-administered Impact of Event Scale-Revised. The relation between S-PTSD and individual vulnerability factors, immediate exposure and post-trauma factors was analysed by gender. RESULTS: S-PTSD was more prevalent in the immediate area than in the peripheral area (women 19% vs 8%; men 8% vs 2%, p<0.01). In the immediate area, S-PTSD was independently associated with birth outside France (men: OR(a) = 13.9, 95% CI 3.7 to 52.8; women: OR(a) = 2.1, 95% CI 1.0 to 4.2), age more than 40 years (men: OR(a) = 4.3, 95% CI 1.01 to 18.2; women: OR(a) = 2.3, 95% CI 1.1 to 4.5), previous psychotropic treatment (men: OR(a) = 11.5, 95% CI 2.4 to 53.6), proximity to the explosion (less educated men only) (OR(a) = 9.3, 95% CI 1.9 to 44.7), rescue efforts (men: OR(a) = 5.2, 95% CI 1.5 to 18.2), temporarily uninhabitable home (men: OR(a) = 5.8, 95% CI 1.9 to 18.1), personal injury (women: OR(a) = 3.7, 95% CI 1.7 to 8.4), financial difficulties (men: OR(a) = 17.4, 95% CI 4.2 to 72.1; women: OR(a) = 3.4, 95% CI 1.7 to 7.1) and inconvenience due to closure of public services (women: OR(a) = 4.1, 95% CI 1.6 to 9.9). CONCLUSIONS: Individual vulnerability, exposure and post-trauma factors were associated with S-PTSD. Vulnerable subgroups, defined by low socioeconomic characteristics may warrant focused screening after such disasters.