In an investigator-blind, randomized cross-over design, male cyclists (mean± SD) age 34.0 (± 10.2) years, body mass 74.6 (±7.9) kg, stature 178.3 (±8.0) cm, peak power output (PPO) 393 (±36) W, and VO2max 62 (±9) ml·kg−1min−1 training for more than 6 hr/wk for more than 3y (n = 20) completed four experimental trials. Each trial consisted of a 2-hr constant load ride at 95% of lactate threshold (185 ± 25W) then a work-matched time trial task (~30min at 70% of PPO). Three commercially available carbohydrate (CHO) beverages, plus a control (water), were administered during the 2-hr ride providing 0, 20, 39, or 64g·hr−1 of CHO at a fluid intake rate of 1L·hr−1. Performance was assessed by time to complete the time trial task, mean power output sustained, and pacing strategy used. Mean task completion time (min:sec ± SD) for 39g·hr−1 (34:19.5 ± 03:07.1, p = .006) and 64g·hr−1 (34:11.3 ± 03:08.5 p = .004) of CHO were significantly faster than control (37:01.9 ± 05:35.0). The mean percentage improvement from control was −6.1% (95% CI: −11.3 to −1.0) and −6.5% (95% CI: −11.7 to −1.4) in the 39 and 64g·hr−1 trials respectively. The 20g·hr−1 (35:17.6 ± 04:16.3) treatment did not reach statistical significance compared with control (p = .126) despite a mean improvement of −3.7% (95% CI −8.8−1.5%). No further differences between CHO trials were reported. No interaction between CHO dose and pacing strategy occurred. 39 and 64g·hr−1 of CHO were similarly effective at improving endurance cycling performance compared with a 0g·hr−1 control in our trained cyclists.