Background We have shown that the reduction of basal insulin (-80%) 40-min before exercise is insufficient to reduce the time spent on hypoglycemia (Roy-Fleming et al., 2018. Diabetes and Metabolism). These results suggest that earlier basal insulin reductions need to be tested. We compared the efficacy of two timings to decrease basal insulin infusion rate to reduce exercise-induced hypoglycemia in patients with T1D using insulin pump therapy. Furthermore, we explored if decreased muscle vasoreactivity (secondary to decreased insulin levels) is associated with a reduced time spent in hypoglycemia. Methods: 13 adults and adolescents (10 adults; 5 adolescents; mean A1C: 8,2±1,0%) practiced 60-min exercise sessions (ergocyle) at 60% VO2peak, 240 minutes after a standardized lunch. In randomized order, we compared an 80% reduction of basal insulin applied 40-min (T-40) and 90-min (T-90) before exercise onset. Near-infrared spectroscopy (NIRS) was used to investigate muscle hemodynamic at vastus lateralis. Venous blood samples for glycemia measurement were drawn every 10 min during exercise. Results: T-90 strategy could reduce hypoglycemia risk during exercise: glycemic drop during exercise tend to be more important during T-40 vs. T-90 strategy (-41.44 ± 57.65 mg/dl vs. -14.05 ± 34.23 mg/dl respectively; p=0.09). This trend is confirmed by the repeated measures ANOVA test, which shows a significant interaction effects (blood glucose level during exercise × strategy “T90 vs. T-40”) (p = 0.01). However, contrary to our hypothesis, the estimation of local muscle perfusion measured by NIRS shows comparable results between 2-strategies. Conclusion: Our preliminary results in 15 DT1 patients (planned 20) show that decreasing basal insulin infusion rate by 80% up to 90 minutes before exercise onset tend to reduce exercise-induced hypoglycemia. This drop does not seem to be related to a decrease in local muscle perfusion.