OBJECTIVES: Confounding issues have not been accommodated adequately in the public health literature of smoking. This study examines the impact of smoking on mortality for patients with diabetes using a propensity score matching approach.  METHODS: We employed the National Health Interview Survey (NHIS) data from 1997 to 2001. Patients who were older than 18 and have been diagnosed with diabetes are included in the study. The study outcome was all cause mortality, where death records obtained from National Death Index were available through Dec. 31, 2006. Propensity score matching (PSM) was performed based upon patients’ smoking status. Time to event model was performed on the PS matched samples. RESULTS: There were 165,057 patients older than 18 participated NHIS survey between 1997 and 2001, of which, 10,194 (6.18%) and had prior diabetes. Our estimation results suggest stop smoking for less than 5 years was associated with a 67% greater risk of death than having never smoked; stop smoking for more than 5 years did not have a different risk of mortality than those who never smoked. In addition, smoking 16-34 cigarettes per day was associated with 52% greater risk of death, and smoking more than 34 cigarettes per day more than doubles the risk relative to the never smokers. CONCLUSIONS: This study indicates smoking reduced survival significantly for patients with diabetes; smoking cessation for more than 5 years prolonged survival. It provides important evidence to decision makers who attempt to advocate smoking cessation programs to patients with diabetes.