Cyclists often use heart rate limits or power output zones, obtained from lactate parameters during incremental exercise testing, to control training intensity. However, the relationship between heart rate or power output, and blood lactate can be changed by several factors including dehydration. Therefore, in the current study we investigated the impact of exercise-induced dehydration on lactate parameters during graded exercise. Nine triathletes completed two test sessions in random order, with a 1-week interval. Each session consisted of 2 graded cycling tests to exhaustion (pretest, posttest), interspersed by a 2-h endurance exercise bout. In one session the cyclists received adequate fluid replacement (EH, 1350 ml . h (-1)) whilst in the other session dehydration was not prevented (DH, 225 ml . h (-1)). Subjects received equal amounts of carbohydrates (150 g) during either condition. The 4-mmol lactate threshold (OBLA) and the d (max) lactate threshold (TH-Dm) were calculated from the power : lactate curves. Weight loss was 0.5 +/- 0.3 kg in EH versus 2.5 +/- 0.2 kg in DH (p < 0.05). Heart rate (HR) at TH-Dm remained unchanged in all test occasions. Conversely, HR at OBLA increased by approximately 10 beats . min (-1) from the pretest to the posttest (p < 0.05), in both EH and DH. Compared to the pretest, in the posttest power output at TH-Dm was reduced (minus approximately 12 %, p < 0.05) in DH, but not in EH. Gross mechanical efficiency at TH-Dm was 20.7 +/- 1 % in the pretest in EH and was not different from the pretest value in DH (21.4 +/- 0.7 %, n.s.). Gross efficiency decreased in the posttest in DH (18.4 +/- 0.6 %, p < 0.05), but not in EH (20.2 +/- 0.8 %, n.s.). It is concluded that heart rate rather than power output should be used to monitor training load in cyclists exercising in environmental conditions predisposing to dehydration. Furthermore, in the latter condition, adequate rehydration is essential to preserve optimal mechanical efficiency.