Background Over the past few decades, there has been a rapid increase in cigarette consumption in China. Large prospective studies are needed to monitor the evolution over the next few decades of the resultant epidemic of tobacco-related deaths in China.

Methods During 2004–2008 the Kadoorie Biobank Study recruited 510 000 adults aged 30–79 from 10 geographically defined regions across China. By Oct 2010, 10 000 deaths had been recorded. Cox regression was used to relate smoking to cause-specific mortality.

Results At baseline, the prevalence of ever smoking regularly was much higher in men (74%) than in women (3%). Among men, the smoking prevalence varied little by area and age but among women, it was associated inversely with year of birth, reflecting mainly the progressive decline in the uptake rate of smoking among younger women over the past few decades. Among men, the overall mortality of ever regular smokers was significantly higher than that of never smokers (adjusted RR 1.32, 95% CI 1.21 to 1.43). There were significant excess mortalities from total cancer, CVD and respiratory disease in male smokers. In urban men, the RR of ever regular smoking was the highest (RR 2.2, 95% CI 2.01 to 2.44) for those who started smoking before age 20 years, suggesting that over 50% of such deaths were caused by smoking.

Conclusion Smoking is already a major cause of premature death in China. The findings among urban men indicate the proportion of male deaths that could eventually be caused by smoking in China as whole.