BACKGROUND: The present study was designed to define the relative contributions of glucose variability and ambient glycemia to the incidence of asymptomatic hypoglycemia in type 2 diabetes. METHODS: Two hundred twenty-two persons with type 2 diabetes were divided into three groups: Group I (n = 53) on insulin sensitizers alone, Group II (n = 87) on oral hypoglycemic agents (OHAs) to include at least one insulin secretagogue, and Group III (n = 82) on insulin alone or in combination with OHAs. Ambient mean glucose concentration (MG) values (in mmol/L) and glycemic variability (SD around the mean glucose value) (in mmol/L) or mean amplitude of glycemic excursions (in mmol/L) were assessed by a continuous glucose monitoring system. Asymptomatic hypoglycemia was recorded over a 48-h period. Poisson regression analysis was used for assessing the potential predictors of hypoglycaemia. RESULTS: The best model fit was obtained with the two following explanatory variables: MG and SD. Hypoglycemia frequency was negatively associated with MG and positively with SD: Log (number of hypoglycemia episodes) = 1.37 - (0.72 x MG) + (1.33 x SD). Odds ratios (95% confidence interval) for hypoglycemic risk were significantly different from 1 for MG at 0.96 (0.95-0.97) (P < 0.0001) and for SD at 1.08 (1.06-1.10) (P < 0.0001). In addition, the risk for hypoglycemia was completely or virtually eliminated when the SD was <1.7 mmol/L irrespective of the ambient glucose level and treatment modality. CONCLUSIONS: As the risk of asymptomatic hypoglycemia increases in the presence of increased glucose variability, avoidance of excess glucose fluctuations should be an important consideration for either reducing or preventing the risk of hypoglycemia in type 2 diabetes.