Many physiological markers vary similarly during training and overtraining. This is the case for the blood lactate concentration ([La-]b), since a right shift of the lactate curve is to be expected in both conditions. We examined the possibility of separating the changes in training from those of overtraining by dividing [La-]b by the rating of perceived exertion ([La-]b/RPE) or by converting [La-]b into a percentage of the peak blood lactate concentration ([La-]b,peak). Ten experienced endurance athletes increased their usual amount of training by 100% within 4 weeks. An incremental test and a time trial were performed before (baseline) and after this period of overtraining, and after 2 weeks of recovery (REC). The [La-]b and RPE were measured during the recovery of each stage of the incremental test. We diagnosed overtraining in seven athletes, using both physiological and psychological criteria. We found a decrease in mean [La-]b,peak from baseline to REC [9.64 (SD 1.17), 8.16 (SD 1.31) and 7.69 (SD 1.84) mmol.l-1, for the three tests, respectively; P < 0.05] and a right shift of the lactate curve. Above 90% of maximal aerobic speed (MAS) there was a decrease of mean [La-]b/RPE from baseline to REC [at 100% of MAS of 105.41 (SD 17.48), 84.61 (SD 12.56) and 81.03 (SD 22.64) arbitrary units, in the three tests, respectively; P < 0.05), but no difference in RPE, its variability accounting for less than 25% of the variability of [La-]b/RPE (r = 0.49). Consequently, [La-]b/RPE provides little additional information compared to [La-]b alone. Expressing [La-]b as a %[La-]b,peak resulted in a suppression of the right shift of the lactate curve, suggesting it was primarily the consequence of a decreased production of lactate by the muscle. Since the right shift of the curve induced by optimal training is a result of improved lactate utilization, the main difference between the two conditions is the decrease of [La-]b,peak during overtraining. We propose retaining it as a marker of overtraining for long duration events, and repeating its measurement after a sufficient period of rest to make the distinction with overreaching.