BACKGROUND: Previous studies showing seasonal clustering of gestational diabetes mellitus (GDM) were conducted in the temperate or frigid zones and mostly included pregnant Caucasian women. This study aims to investigate the association of ambient temperature with prevalence of GDM in Taiwan, a sub-tropic country. METHODS: This population-based cohort study comprised women (nÊ=Ê371,131) who gave births between 2013 and 2014; of which, 43,538 (11.7%) were diagnosed with GDM. The mean daily temperature and difference in temperature within a day was calculated over a 35-day period prior to GDM diagnosis or the first day of the 27th gestational week (for non-GDM subjects). Multiple logistic regression models with generalized estimation equation were performed to estimate the adjusted odds ratio (aOR) and 95% confidence interval (CI) of GDM in association with temperature. RESULTS: After controlling for potential confounders, summer and fall were associated with higher risk of GDM diagnosis, with aOR [95% CI] of 1.05 [1.04-1.07] and 1.04 [1.02-1.06] in reference to winter. Additionally, an increase of 1Ê¡C from 14Ê¡C to 27Ê¡C was associated with an aOR of 1.03 [1.02-1.03]. The aOR greatly increased to 1.54 [1.48-1.60] after 28Ê¡C. An increase of 1Ê¡C difference within a day was associated with a reduced aOR at 0.90 [0.87-0.92]. CONCLUSION: A higher prevalence of GDM was associated with a higher daily temperature, but with a smaller difference in temperature within a day.