This study aimed to analyse the relationship between smoking and preterm birth (22–36 weeks) according to the main obstetric complications leading to the preterm birth, both overall and by parity. The EUROPOP study is a case-control study carried out between 1994 and 1997; 3,787 preterm and 5,602 full-term births were included, from maternity units in 10 countries, using the same protocol. Social, demographic and medical information was collected after delivery, from obstetric records and interviews with the mothers. Cases were classified according to the main obstetric complication (hypertension, haemorrhage, preterm premature rupture of membranes (PPROM), idiopathic spontaneous preterm labour, intrauterine growth retardation, all other causes). Multiple logistic regression analysis was used to control for confounders. Twenty four percent of cases and 20% of controls were smokers. Smoking during pregnancy, heavy smoking (≥10 cigarettes per day) in particular, was a risk factor for preterm birth (aOR = 1.39, 95% CI:1.20–1.60). Smoking increased the risk of preterm delivery due to all obstetric complications other than hypertension. For these complications, the risk of preterm delivery associated with smoking was higher for multiparae (aOR = 1.46, 95% CI:1.24–1.71) than for primiparae (aOR = 1.18, 95% CI:1.00–1.38). In conclusion, smoking during pregnancy increases the risk of preterm birth among women with all obstetric complications except hypertension. This association is stronger in multiparae than in primiparae and the risk is higher for heavy smokers.