Background: Residing in greener places may be protective against diabetes mellitus (DM) but evidence is scarce and comes mainly from developed countries.Objectives: To investigate associations of residential greenness with DM prevalence and glucose-homeostasis markers in Chinese adults and whether these associations were mediated by air pollution, physical activity, and body mass index.Methods: In 2009, a total of 15,477 adults from the cross-sectional 33 Communities Chinese Health Study provided blood samples and completed a questionnaire. We considered fasting and 2-h glucose and insulin concentrations, as well as the homoeostasis model assessment of insulin resistance and beta-cell function, as glucose-homeostasis markers. DM was defined according to the American Diabetes Association's recommendations. Residential greenness was estimated by two satellite-derived vegetation indexes - Normalized Difference Vegetation Index (NDVI) and Soil Adjusted Vegetation Index (SAVI). Nitrogen dioxide and particulate matter <= 2.5 mu m were used as air pollution proxies. Associations were assessed by two-level adjusted logistic and linear regression models.Results: A 0.1-unit increase in NDVI500 m and SAVI(500 m) was significantly associated with lower odds of DM by factors of 0.88 (95% Confidence Interval 0.82-0.94) and 0.80 (0.72-0.90), respectively. Higher greenness was also significantly associated with lower fasting and 2-h glucose levels, 2-h insulin level, as well as lower insulin resistance and higher beta-cell function. Air pollution and body mass index significantly mediated 6.9-51.1% and 8.6-78.7% these associations, respectively, while no mediation role was observed for physical activity.Conclusions: Higher residential greenness appears to be associated with a lower prevalence of DM. This association might be due to glucose and insulin metabolism and pancreatic beta-cell function. Lower levels of air pollution and body mass index can be pathways linking greenspace to diabetes.