Operational Modal Analysis (OMA) allows the structural identification of systems in working conditions, moving from output data only. External forces remain not measured. Basically some fundamental assumptions have to be satisfied: (i) the not known loads acting on the system need to have the form of white noise sequences, (ii) in the case of multi-point excitation the external inputs are required to be strictly uncorrelated. In the field of vehicle dynamics, OMA can be utilised to assess the performance of different suspension systems equipping the same type of vehicle. The output data recorded at the different sensor locations during vehicle road tests are affected by a certain correlation among the road forces acting on the wheels, mainly that loading the front and the rear axle, even if inputs on wheels belonging to the left and the right side could be also in some way correlated. In this paper the correlation between the road inputs acting on the front and the rear axle of a passenger car has been investigated with the aim of obtaining a novel OMA formulation moving from the expressions of the modal decomposition of the output power spectral densities that hold in this case.