OBJECTIVE: To describe the rates of and risk factors associated with iatrogenic and spontaneous preterm birth and the variation in rates between hospitals. DESIGN: Cohort study using electronic health records. SETTING: English National Health Service. POPULATION: Singleton births between 1 April 2015 and 31 March 2017. METHODS: Multivariable Poisson regression models were used to estimate adjusted risk ratios (adjRR) to measure association with maternal demographic and clinical risk factors. MAIN OUTCOME MEASURES: Preterm births (<37 weeks of gestation) were defined as iatrogenic or spontaneous according to mode of onset of labour. RESULTS: Of the births, 6.1% were preterm and of these, 52.8% were iatrogenic. The proportion of preterm births that were iatrogenic increased after 32 weeks. Both sub-groups were associated with previous preterm birth, extremes of maternal age, socio-economic deprivation and smoking. Iatrogenic preterm birth was associated with higher body mass index (BMI) (BMI >40 kg/m2 adjRR 1.59, 95% CI 1.50-1.69) and previous caesarean (adjRR 1.88, 95% CI 1.83-1.95). Spontaneous preterm birth was less common in women with a higher BMI (BMI >40 kg/m2 adjRR 0.77, 95% CI 0.70-0.84) and in women with a previous caesarean (adjRR 0.87, 95% CI 0.83-0.90). More variation between NHS hospital trusts was observed in rates of iatrogenic, compared with spontaneous, preterm births. CONCLUSIONS: Just over half of all preterm births resulted from iatrogenic intervention. Iatrogenic births have overlapping but different patterns of maternal demographic and clinical risk factors to spontaneous preterm births. Iatrogenic and spontaneous sub-groups should therefore be measured and monitored separately, as well as in aggregate, to facilitate different prevention strategies. This is feasible using routinely acquired hospital data.