Aims. - To examine the seasonality and effects of temperature on levels of fasting plasma glucose (FPG). Methods. - We collected health data from the Kailuan cohort study. FPG, blood pressure and individual information including age, gender, body mass index, smoking status, drinking habit, physical activities, income, work type, education level, and history of diabetes, were collected for each participant. Daily weather conditions were collected during the study period of 2006-2011. A total of 49,417 participants who had three times of health examination were included to the analyses. Generalized additive mixed models were used to examine the effects of temperature and seasonality on FPG levels, while controlling for potential confounders. Results. - FPG level was higher in winter and spring than that in autumn and summer. For all participants, the FPG winter level increased 0.31 mmol/L [95% confidence interval (CI), 0.28-0.33 mmol/L] in comparison with autumn. The association between temperature and FPG levels was U-shaped. For all participants, the change in FPG levels associated with extreme cold temperature (-6.7 degrees C), moderate cold temperature (2.4 degrees C), moderate hot temperature (23.7 degrees C), and with extreme hot temperature (28.1 degrees C), in comparison with threshold (18.1 degrees C) were 0.12 mmol/L (95% CI: 0.10-0.14 mmol/L), 0.10 (95% CI: 0.09-0.12 mmol/L), 0.06 (95% CI: 0.04-0.08 mmol/L), and 0.12 mmol/L (95% CI: 0.08-0.16 mmol/L), respectively. Conclusion. The findings suggest that there may be strong relationships between FPG levels and season and ambient temperature. In particular, there were higher FPG levels in the winter and at extreme cold and hot temperatures. (C) 2016 Elsevier Masson SAS. All rights reserved.