Hypoglycemia is the most feared side effect of diabetes therapy with blood glucose-lowering agents. The fear of hypoglycemia often contributes to poor metabolic control of patients with diabetes. Therefore, integration of a hypoglycemia warning signal into continuous glucose monitoring systems represents an important additional help for patients with diabetes. The warning signal can be triggered at a preset level based on the current glucose values (as provided with the presently available glucose monitoring systems) or on prospective trend analysis offering the possibility to predict the risk of a hypoglycemic event in an anticipatory manner. Using the approach of a "Finite State Machine," such a more advanced warning system can completely be described as a finite collection of four states and possible transitions in-between. Most of the currently available glucose monitoring systems measure glucose in the interstitial fluid (ISF) of the dermal or subcutaneous tissue but are calibrated to blood glucose levels. This requires a number of factors to be taken into account: precision and accuracy of the glucose measurements, physiological and physical lag time, and calibration of the glucose monitoring system. From our point of view, the analytical performance of the system should be such that the majority of all hypoglycemic episodes are correctly diagnosed (>75%). Inconsistent findings regarding physiological discrepancies between blood and ISF glucose, which usually are described as a physiological lag time, range from some seconds up to 15 min. They can be observed especially during dynamic blood glucose changes (>3 mg/dL/min) and may represent major challenges for the development of a reliable hypoglycemia warning signal. In addition to possible physiological time lags, device-inherent physical lag times must be considered when selecting the threshold for the warning signal. Despite these problems, most probably all patients with diabetes who are treated with blood glucose-lowering agents will benefit from such a system since their safety and quality of life can be greatly improved, including an optimized metabolic control and lowered diabetes-related mortality. The benefit will be greatest for patients with hypoglycemia unawareness or impaired perception of hypoglycemic symptoms. The risks related to the use of a hypoglycemia warning signal seem to be low if certain precautionary measures are taken. In any case, additional clinical-experimental studies in healthy subjects as well as long-term clinical studies in diabetic patients are necessary to further evaluate the efficacy, benefits, and risks of different hypoglycemia warning concepts implemented in the different continuous glucose monitoring systems.