Objective The activity of brown adipose tissue is sensitive to changes in ambient temperature. A lower exposure to cold could result in an increased risk of developing diabetes at population level, although this factor has not yet been sufficiently studied. Design We studied 5072 subjects, participants in a national, cross-sectional population-based study representative of the Spanish adult population (Di@bet.es study). All subjects underwent a clinical, demographic and lifestyle survey, a physical examination and blood sampling (75 g oral glucose tolerance test). Insulin resistance was estimated with the homeostasis model assessment (HOMA-IR). The mean annual temperature (°C) in each individual municipality was collected from the Spanish National Meteorology Agency. Results Linear regression analysis showed a significant positive association between mean annual temperature and fasting plasma glucose (β: 0.087, P < 0.001), 2 h plasma glucose (β: 0.049, P = 0.008) and HOMA-IR (β: 0.046, P = 0.008) in multivariate adjusted models. Logistic regression analyses controlled by multiple socio-demographic variables, lifestyle, adiposity (BMI) and geographical elevation showed increasing odds ratios for prediabetes (WHO 1999), ORs 1, 1.26 (0.95–1.66), 1.08 (0.81–1.44) and 1.37 (1.01–1.85) P for trend = 0.086, diabetes (WHO 1999) ORs 1, 1.05 (0.79–1.39), 1.20 (0.91–1.59) and 1.39 (1.02–1.90) P = 0.037, and insulin resistance (HOMA-IR ≥75th percentile of the non-diabetic population): ORs 1, 1.03 (0.82–1.30), 1.22 (0.96–1.55), 1.26 (0.98–1.63) (P for trend = 0.046) as the mean annual temperature (into quartiles) rose. Conclusions Our study reports an association between ambient temperature and the prevalence of dysglycemia and insulin resistance in Spanish adults, consistent with the hypothesis that a lower exposure to cold could be associated with a higher risk of metabolic derangements.