The rise of output-only modal analysis has offered an economical and efficient way to identify modal parameters of civil engineering structures, namely natural frequencies, damping ratios and mode shapes. However, since the forcing is unknown, it is not possible to directly estimate modal masses, and estimates of damping ratios may be inaccurate. With the advancement of wireless sensor networks both vehicle and bridge responses can be simultaneously measured. This offers the possibility of estimating true Frequency Response Functions (FRFs), since the vehicle acceleration gives an estimate of the force input to the bridge. Hence in principle it is possible to estimate modal masses and more accurate damping ratios. However, the spatial and temporal variation of the moving load from a passing vehicle gives challenges to this idea and precludes the direct use of existing single-input-multiple-output (SIMO) system identification methods. Even if the system is treated as a multiple-input-multiple-output (MIMO) one, the inputs are highly correlated so existing methods for these systems are not applicable either. For this reason, a two-stage strategy is proposed to modify an existing method to solve this moving load problem.