The purpose of this paper was to estimate the road profile irregularities that affect vehicles. The studied random road profiles are related to the ISO 8608 standard. The identification of these disturbances is very important to select the adequate control law for an active suspension. Many studies have been carried out in this field but they were either costly because their road profile measurements were achieved using longitudinal profile analyser or laser sensors or needed long computation time particularly the neural network based models. This study proposed the implementation of a new fast and simple technique, called the Independent Component Analysis (ICA) to identify the road profile, based on the so-called inverse problem. Knowing the dynamic responses of the system (observed signals), the ICA allows the identification of road excitation. These responses can be either measured via sensors or computed numerically. To this end, three models were studied: a full car model, a half car (pitch-bounce and roll model) and a quarter car. The full car is considered as the real model. So the dynamic responses, which make up the ICA observed signals, are computed from this model. In the second part of this paper, we studied the ICA efficiency to construct the estimated road profile. The obtained results were validated using some performance criteria and the robustness of the method was assessed using the sprung mass variation and noise effect. The obtained results show that the ICA can identify the road disturbances adequately.