Introduction: System identification of the neuromuscular controller that regulates human balance, gives insight in the causes and effects of human balance disorders. The combination of varying perturbation properties in literature and lack of rationale behind chosen perturbation properties, reveals possible violations of assumptions made to perform system identification. This study investigates the effect of number of repetitions of the perturbation signal and the perturbation amplitude on the identification quality. Methods: 12 subjects were perturbed at their support surface with a multisine signal. Kinematics of the subjects were recorded with a motion capture system; ground reaction forces were recorded with force plates. The best linear approximation of the neuromuscular controller was identified. Measures for variability of identification with respect to the number of repetitions and for nonlinearity with respect to the perturbation amplitude were calculated. Results: Identification variability was found to significantly decrease with number of repetitions. Nonlinearity within the neuromuscular controller increased significantly with perturbation amplitude. Conclusions: After having measured 6-8 repetitions, measuring subsequent repetitions does not decrease identification variability significantly for the following 6-7 repetitions. The balance between suppressing noise effects by using a high amplitude perturbation and minimising nonlinear effects by using a low amplitude perturbation was identified to be optimized between a perturbation signal amplitude of 8 cm PtP and 11 cm PtP.