This study aimed to investigate the relationship of ankle-brachial index (ABI) with all-cause mortality and cardiovascular (CV) mortality in Chinese inpatients after 3 years of follow-up. All the participants were inpatients at high risk of atherosclerosis and enrolled from July to November 2004. A total of 3210 patients were followed up until an end point was reached or until February 2008. The mean follow-up time was 37.65+/-1.54 months. Patients with ABI <or=0.4 were significantly older than other ABI categories (P<0.001) at baseline. The all-cause and CV disease (CVD) mortality were highest (37.7 and 27.5%, respectively) after a 3-year follow-up in patients with ABI <or=0.4. There was a significant increasing tendency in mortality with decreasing ABI levels (P<0.001). The Kaplan-Meier curves of survival showed a decreasing survival rate with the ABI decreasing, not only for all-cause mortality but also for CVD mortality (P<0.001). Adjusted for other risk factors, patients with ABI <or=0.4 were 3.105 times (95% CI 1.936-4.979) as likely to die as those with ABI in the range of 1.00-1.4; patients with ABI <or=0.4 were about five times (95% CI 2.740-8.388) as likely to die of CVD as those with ABI in the range of 1.00-1.4. Even patients with ABI in the range of 0.41-0.90 were more than 1.5 times as likely to die (relative risk=1.534, 95% CI 1.199-1.962), or die of CVD (relative risk=2.031, 95% CI 1.479-2.789) as those with ABI in the range of 1.00-1.4. Low ABI is related to a higher all-cause and CV mortality compared with normal ABI. ABI as a marker of atherosclerosis should be routinely evaluated to assess the risk of CV mortality in Chinese patients.