Given that Thoroughbred horses' canter is an asymmetric gait, not only speed but also leading or trailing limbs could affect muscle activities. However, the muscle activity during a canter remains poorly understood. Hence, we aimed to investigate speed and lead-side (leading or trailing) effects on surface electromyography (sEMG) during a canter. The sEMG data were recorded from left Musculus brachiocephalicus (Br), M. infraspinatus (Inf), long head of M. triceps brachii (TB), M. gluteus medius (GM), M. semitendinosus (ST), and M. flexor digitorum longus of seven Thoroughbreds with hoof-strain gauges at the left hooves. Horses cantered on a flat treadmill at 7, 10, and 13 m/s for 25 s each without lead change. Subsequently, the horses trotted for 3 min and cantered at the same speed and duration in the opposite lead side ("leading" at the left lead and "trailing" at the right lead). The order of the lead side and speed was randomized. The mean of 10 consecutive stride durations, duty factors, integrated-EMG values (iEMG) for a stride, and muscle onset and offset timing were compared using a generalized mixed model (P < 0.05). Stride durations and duty factors significantly decreased with speed regardless of the lead side. In all muscles, iEMG at 13 m/s significantly increased compared with 7 m/s (ranging from +15% to +134%). The lead-side effect was noted in the iEMG of Br (leading > trailing, +47%), Inf (leading > trailing, +19%), GM (leading < trailing, +20%), and ST (leading < trailing, +19%). In TB, GM, and ST, muscle onset in trailing was earlier than the leading, while offset in the leading was earlier in Br. In conclusion, different muscles have different responses to speed and lead side; thus, both the lead side and running speed should be considered during training and/or rehabilitation including canter or gallop.