Kinetic parameters for pulmonary O2 uptake at exercise onset are estimated by non-linear regression on repeated responses assembled together. The native data contain the information, thus the "stacking" of the responses should provide correct values and uncertainties. Kinetic parameters and corresponding uncertainties (computed from the asymptotic standard errors; ASE) were estimated on 104 simulated noisy responses (with time constant τ=25s), repeated 10 times and assembled over an increasing number of repetitions (Nr) by "stacking" or ensemble averaging the responses processed to obtain 1s bins ("1-s-bins"). Independent of the assembling method, the average estimated τ amounted to ~25.05s. Independent of Nr, the "1-s-bins" and the "stacking" yielded an ASE/SD ratio for τ amounting to ~0.52 and ~0.98, respectively, resulting in a probability of including τ=25s within the estimated uncertainty from an individual kinetics amounting to ~70% and >94% for the "1-s-bins" and the "stacking", respectively. In conclusion, the "stacking" allowed obtaining, also for individual kinetics, coherent estimated parameters and associated uncertainties
