A crucial step in developing a glucose monitoring system using a subcutaneous implanted glucose sensor is the transformation of the sensor signal (a current) into an estimation of a blood glucose concentration. We have developed an Electronic Control Unit (ECU) able to recognize, before and after a glucose load, that the sensor current presents a plateau, thus triggering an alarm asking for blood glucose determination. The system, fed with these results, subsequently transforms the current into an estimation of glucose concentration by linear extrapolation based on the sensor sensitivity and the background current computed from the two sets of current and glycaemia values (two-point calibration). In addition, the system is able to trigger an alarm when this estimation decreases below a threshold that can be set by the user. This system was evaluated in experiments performed in 12 normal rats. The quality of the calibration was assessed by comparing, by error grid analysis, the data displayed on the liquid-crystal display of the ECU to concomitant plasma glucose concentration determined at frequent intervals, 65 +/- 6 and 26 +/- 5% of the values were in zones A (good) and B (acceptable estimation) of the grid, respectively. The system was set to trigger an alarm when the estimation of glucose concentration decreased below 70 mg/dl. Following an insulin administration, the alarm was triggered when the system displayed a 64 +/- 2 mg/dl glucose concentration. The concomitant plasma glucose concentration was 59 +/- 5 mg/dl (NS). In conclusion, this work validates experimentally the new, user-friendly method for calibrating the glucose sensor integrated into the ECU, based on an automatic detection of plateaus. The quality of the sensor calibration performed with this procedure is compatible with the appropriate functioning of this continuous glucose monitoring system, which was demonstrated by its ability to detect mild hypoglycaemia following insulin injection.