Background: Postnatal depression can have a significant impact on both maternal and child health. There have been very few studies undertaken in the developing world to identify women at risk of postnatal depression. The present study aimed to investigate risk factors for the occurrence of postnatal depression in urban South African women. Method: A large prospective population cohort study identified antenatal risk factors for postnatal depression in Soweto, Johannesburg, South Africa. One thousand and thirty-five women were interviewed in the antenatal period and subsequently completed the Pitt Depression Questionnaire in the postnatal period. Results: 170/1035 (16.4%) women were probable cases of postnatal depression. The strongest independent predictors of postnatal depression were exposure to extreme societal stressors (witnessing a violent crime/danger of being killed) (adjusted Odds Ratio 2.468 (95% Confidence Interval 1.509, 4.037)) and reporting difficulties with their partner (adjusted OR 1.645 (1.088, 2.488)). Limitations: There was some loss of the sample during follow-up (35.1%), which was to be expected given the turbulent nature of the study setting. The measures were questionnaires administered by interview. Conclusion: Postnatal depression is common in women in developing world settings, such as this part of Southern Africa. Although some of the risk factors for postnatal depression were similar to those identified in studies in developed nations, some important differences exist, most notably antenatal exposure to extreme societal stressors. This study shows that it is possible to identify women at risk during pregnancy in a developing world setting.