This paper describes a case study of identifying the physical model (or the grey box model) of a hydraulic test robot. The obtained model is intended to provide a basis for model-based control of the robot. The physical model is formulated in continuous time and is derived by application of the laws of physics on the system. The unknown (or uncertain) parameters are estimated with Maximum Likelihood (ML) parameter estimation. The identified model has been evaluated by comparing the measurements with simulation of the model. The identified model was much more capable of describing the dynamics of the system than the deterministic model.