Any variation of the test protocol for incremental testing (IT) in cycle ergometry may affect the accuracy of the determination of anaerobic threshold (AnT). For lactate threshold concepts, indicating the maximum lactate steady-state (max Lass), the formation of the quasi-steady-state (QSS=95% of steady-state level) is evident. Previous studies have not specified the time that it takes for blood lactate to stabilize following incremental changes in WL. The purpose of this study was to identify the minimum duration of exercise necessary to establish QSS following various increments in WL (10, 20, 30, 40 and 50 W). Eight male endurance-trained cyclists [relative maximal oxygen consumption = 64.8 (4.2) ml x kg(-1) x min(-1)] performed three different exercise tests on a cycle ergometer: (1) an exhaustive IT with a starting WL of 100 W, followed by 20-W increments every 3 min; (2) a threshold test with 20-W increments every 9 min to determine the MaxLaSS; and (3) five incremental exercise tests (from 100/110 W, with 20-W increments every 3 min) with a final 10-, 20-, 30-, 40- or 50- W increment lasting 10 min, at 10 W below MaxLaSS (T10-T50 experiments). The time constant of lactate kinetics (tau), the time constant of lactate elimination, and the time taken to elicite QSS, defined as 95% of the time taken to reach steady-state level (t95%), were calculated in the T10-T50 experiments. The tau and t95% increased significantly with WL increment size: the correlation was not linear. Smaller WL increments required proportionally longer durations. Mean (SD) t95% values (min:s) were 1:57 (0:27) (T10), 2:58 (0.16) (T20), 4:08 (0:23) (T30), 4:45 (0:45) (T40) and 5:06 (0:43) (T50). The application of these references in IT protocols may lead to an extension of total test duration, particularly with smaller increments. Therefore, lactate threshold modelling, the training status of the athletes and comparability with lactate measurements obtained during training events should be considered. IT protocols not accomplishing QSS criteria may effect an underestimation of WL-related lactate values and an overestimation of lactate thresholds, which indicate MaxLaSS, especially in highly trained athletes. This suggests that the establishment of an increment-size-dependent t95% may reduce protocol-related influences on AnT and standardize the use of the AnT in IT procedures in the training management of elite cyclists.