Reducing pre-exercise exogenous insulin may be an effective way to combat hypoglycaemia in exercising people with type 1 diabetes (T1DM; Rabasa-Lhoret et al., 2001; Mauvais-Jarvis et al., 2003). However there are conflicting views on the best insulin reduction strategy to perform prior to exercise to preserve blood glucose after exercise. This study examined changes in insulin isoforms (Lispro/Aspart) and blood glucose in T1DM individuals before, during and after exercise following reductions in self-administered insulin dose. With ethical approval, eight T1DM individuals (7 males, 1 female, 34±13 years, 84±16 kg, Hb_A1c 8.3±0.6%, VO_2peak 44±7 ml.kg^-1.min^-1) volunteered for this study. In randomised order and on separate occasions after an overnight fast, participants self-administered 100%, 75%, 50% or 25% of their rapid-acting insulin dose subcutaneously into the abdomen prior to consuming a 1121 kJ meal (60g carbohydrates, 2g protein, 2g fat). After a 2 h rest, participants performed 45 min of treadmill running at 70±1% VO_2peak. Blood glucose (BG) and insulin were measured for 2 h pre and 3 h post-exercise. Data were analysed using repeated-measures ANOVA and presented as mean±SD. Results are expressed as changes from resting values. In the 2 h before exercise peak insulin values under 100% were greater than 50% and 25% (P<0.05). Consequently, there was a trend for lower peak BG under 100% (6.1±2.6 mM.l^-1) than 75% (7.6±2.3 mM.l^-1, P=0.09), 50% (7.2±3.1 mM.l^-1, P=0.09) and 25% (8.5±4.2 mM.l^-1, P=0.05). After 2 h, insulin under 100% (112±97 nM.l^-1) tended to be greater than 75% (86±65 nM.l^-1, P=0.08), 50% (56±47 nM.l^-1, P=0.09) and 25% (70±55 nM.l^-1, P=0.06) with BG under 100% (3.9±2.7 mM.l^-1) lower than 75% (6.5±2.4 mM.l^-1, P=0.01), and tending to be lower than 50% (5.9±3.0 mM.l^-1, P=0.07), and 25% (7.4±4.6 mM.l^-1, P=0.06). BG decreased under all conditions following exercise with the reduction under 100% (6.3±3.1 mM.l^-1) similar to 75% and 50% and greater than 25% (4.0±3.3 mM.l^-1, P<0.05). Insulin concentrations did not change after exercise during 100% and 75% but tended to increase after 25% and 50% (P=0.07). In the 3 h following exercise, insulin decreased to resting values under all conditions (100% 15±61, 75% 8±76, 50% 7±76, 25% 22±37 nM.l^-1, P<0.05) with the drop in insulin under 100% greater than 25% (P<0.05). Although BG did not change over 3 h post-exercise, the decline from rest was greater under 100% (-2.5±5.2 mM.l^-1) compared with 75% and 25% (0.4±5.3 and 3.6±4.2 mM.l^-1, respectively, P<0.05) but similar to 50% (-1.5±6.0 mM.l^-1, NS). The findings demonstrate a 75% reduction to rapid-acting insulin two hours before exercise best preserves blood glucose concentrations during and after aerobic exercise in T1DM individuals and may offer a useful strategy in the prevention of hypoglycaemia.