Smoking is a risk factor for intra-operative pulmonary complications and a wide range of postoperative pulmonary, cardiovascular, infection and wound-related complications. These may all lead to unplanned postoperative intensive care admission. We tested the hypothesis that smokers have an increased incidence of postoperative intensive care admission and more postoperative complications than nonsmokers in a general and orthopaedic surgical population. The following information was assessed in 6026 surgical patients: age, sex and smoking status (pack-years), history of heart and lung disease, American Society of Anesthesiologists (ASA) physical classification, intensive care admission and postoperative complications. Two thousand five hundred and twenty-six (46%) were smokers but for 620 patients (10.3%) smoking status was not confirmed. Postoperative intensive care admission was required by 319 patients (5.3%). Patients with > 50 pack-years were admitted to the intensive care more frequently than were smokers with < or = 50 pack-years history and nonsmokers (p < 0.001). Ex-smokers with > 50 pack-years history had the same risk of postoperative admission to intensive care as smokers with > 50 pack-years history. Smokers admitted to intensive care with > 50 pack-years history had a higher incidence of chronic lung disease (p < 0.005) and heavy alcohol consumption (p < 0.001). These smokers also had a higher incidence of postoperative pulmonary complications (odds ratio = 3.91, p < 0.01). The mortality rate was 37% in smokers with > 50 pack-years history and 24% in nonsmokers (odds ratio = 2.02, p = 0.08). We conclude long-term tobacco smoking (> 50 pack-years) carries a higher risk of postoperative admission to intensive care, and there seems to be a dose relationship between the amount of tobacco consumed and the risk of postoperative intensive care admission.