Objective. This population study was undertaken to determine whether previous abortion is an independent risk factor for pre-term birth and to calculate population-attributable risks for risk factors. Methods. All South Australian first singleton births in 1998-2003 (n = 42 269) were included in a multivariable logistic regression analysis, comparing pre-term births with term births. Results. Risk factors for pre-term birth were found to be: being indigenous, single, a smoker [adjusted odds ratio (AOR) 1.28, 95% confidence interval 1.17-1.41], age 40 years or older, reproductive technology assistance, threatened miscarriage, antepartum haemorrhage, urinary tract infection, pregnancy hypertension and suspected intra-uterine growth restriction. A previous spontaneous abortion was of borderline statistical significance, whereas a previous induced abortion (AOR 1.25, 1.13-1.40) was an independent risk factor. A dose-response relationship was found with increasing number of previous spontaneous or induced abortions. Population-attributable risks were highest for pregnancy hypertension (12.4%) and antepartum haemorrhage (9.2%). Smoking and previous induced abortion had risks of 4.7% and 2.7%, respectively. Among indigenous women, 51% of whom smoked, 16.4% of pre-term birth could be attributed to smoking. Conclusions. A previous induced abortion and smoking during pregnancy (particularly among indigenous women) are preventable risk factors for pre-term birth. Their population-attributable risks are likely to be under-estimates from under-reporting. ; Rosanne Freak-Poli, Annabelle Chan, Graeme Tucker and Jackie Street