Background and aim: We sought to determine the effect of regression to normal glucose tolerance (NGT) or progression to diabetes in early years of impaired glucose tolerance (IGT) on subsequent risk of stroke. Methods: In 1986, 576 adults aged 25 years and older with impaired glucose tolerance in Da Qing, China, were randomly assigned by clinic to control, diet, exercise, or diet plus exercise intervention groups for a six-year period. Subsequently participants received medical care in their local clinics. We tracked participants for additional 17 years to ascertain stroke events and other outcomes. Results: At the end of 6-year intervention trial follow-up, 272 (50.2%) had progressed to diabetes, 169 (31.2%) regressed to normal glucose tolerance, and 101 (18.6%) remained impaired glucose tolerance. During the subsequent 17-year follow-up, 173 (31.9%) developed a stroke, 26.7% of normal glucose tolerances, 30.7% of impaired glucose tolerances, and 36.1% of those with diabetes. After controlling for age, sex, baseline blood pressure, smoking, total cholesterol, previous cardiovascular disease and intervention group, those who developed diabetes in the first six years had a higher incidence of stroke than those who reverted to normal glucose tolerance (HR = 1.49, 95% CI 1.01-2.19, p = 0.04), whereas for those who remained impaired glucose tolerance compared to those who regressed to normal glucose tolerance the HR was 1.25 (95% CI 0.80-1.93; p = 0.30). A 1-mmol/L increase in both fasting and 2-h post-load plasma glucose from entry to end of the six-year trial was significantly associated with a higher risk of development of stroke in the subsequent 17 years, respectively (HR = 1.07, 95% CI 1.03-1.11, p<0.0001 for fasting glucose, HR = 1.05, 95% CI 1.02-1.09, p = 0.007 for 2-h post-load plasma glucose). Conclusions: Among Chinese adults with impaired glucose tolerance, early progression to diabetes predicted a higher risk of stroke, compared those who regressed to normal glucose tolerance.