OBJECTIVE Continuous glucose monitoring system (CGMS) provides detailed information on glucose fluctuations. The aim was to establish whether CGMS could be used during physical exercise and whether it detects more episodes of hypoglycaemia and hyperglycaemia than frequent blood glucose measurements. METHODS Adolescents with type 1 diabetes (12 girls and 47 boys) participated in three annual sports camps that lasted for 3-4 days and included different types of exercise: soccer, floorball + cross-country skiing and golf. During the study, blood glucose values, mean 8.7 ± 3.3 per day, were obtained with Hemocue in parallel with the CGMS. RESULTS Ninety-eight per cent of the participants used the sensor at all times during the camps. Eighty-seven per cent of the sensors gave adequate signals for 24 h and 66% for 48 h. Median durations of hypoglycaemia and hyperglycaemia were 1.7 h per day and 3.8 h per day, respectively. The CGMS identified significantly more episodes of hypoglycaemia (p < 0.005) and hyperglycaemia (p < 0.005) during the day and night than frequent blood glucose tests. CONCLUSIONS We demonstrate that, even during days that included episodic strenuous physical exercise, CGMS could provide useful information on glucose fluctuations during day and night, albeit with significant failure rates.