PURPOSE: For wheelchair users with a spinal cord injury, the lower body may be a more convenient cooling site than the upper body. However, it remains unknown if leg cooling reduces thermal strain in these individuals. We compared the impact of upper-body versus lower-body cooling on physiological and perceptual outcomes during submaximal arm-crank exercise under heat stress in individuals with paraplegia. METHODS: Twelve male participants with paraplegia (T4-L2, 50% complete lesion) performed a maximal exercise test in temperate conditions, and three heat stress tests (32°C, 40% relative humidity) in which they received upper-body cooling (COOL-UB), lower-body cooling (COOL-LB), or no cooling (CON) in a randomized counterbalanced order. Each heat stress test consisted of four exercise blocks of 15 min at 50% of peak power output, with 3 min of rest in between. Cooling was applied using water-perfused pads, with 14.8-m tubing in both COOL-UB and COOL-LB. RESULTS: Gastrointestinal temperature was 0.2°C (95% confidence interval (CI), 0.1°C to 0.3°C) lower during exercise in COOL-UB versus CON (37.5°C ± 0.4°C vs 37.7°C ± 0.3°C, P = 0.009), with no difference between COOL-LB and CON ( P = 1.0). Heart rate was lower in both COOL-UB (-7 bpm; 95% CI, -11 to -3 bpm; P = 0.01) and COOL-LB (-5 bpm; 95% CI, -9 to -1 bpm; P = 0.049) compared with CON. The skin temperature reduction at the cooled skin sites was larger in COOL-LB (-10.8°C ± 1.1°C) than in COOL-UB (-6.7°C ± 1.4°C, P < 0.001), which limited the cooling capacity in COOL-LB. Thermal sensation of the cooled skin sites was improved and overall thermal discomfort was lower in COOL-UB ( P = 0.01 and P = 0.04) but not in COOL-LB ( P = 0.17 and P = 0.59) compared with CON. CONCLUSIONS: Upper-body cooling more effectively reduced thermal strain than lower-body cooling in individuals with paraplegia, as it induced greater thermophysiological and perceptual benefits.