We present a method to recover mass profiles of galaxy clusters by combining data on thermal Sunyaev-Zeldovich (tSZ) and X-ray imaging, thereby avoiding to use any information on X-ray spectroscopy. This method, which represents a development of the geometrical deprojection technique presented in Ameglio et al. (2007), implements the solution of the hydrostatic equilibrium equation. In order to quantify the efficiency of our mass reconstructions, we apply our technique to a set of hydrodynamical simulations of galaxy clusters. We propose two versions of our method of mass reconstruction. Method 1 is completely model-independent, while Method 2 assumes instead the analytic mass profile proposed by Navarro et al. (1997) (NFW). We find that the main source of bias in recovering the mass profiles is due to deviations from hydrostatic equilibrium, which cause an underestimate of the mass of about 10 per cent at r_500 and up to 20 per cent at the virial radius. Method 1 provides a reconstructed mass which is biased low by about 10 per cent, with a 20 per cent scatter, with respect to the true mass profiles. Method 2 proves to be more stable, reducing the scatter to 10 per cent, but with a larger bias of 20 per cent, mainly induced by the deviations from equilibrium in the outskirts. To better understand the results of Method 2, we check how well it allows to recover the relation between mass and concentration parameter. When analyzing the 3D mass profiles we find that including in the fit the inner 5 per cent of the virial radius biases high the halo concentration. Also, at a fixed mass, hotter clusters tend to have larger concentration. Our procedure recovers the concentration parameter essentially unbiased but with a scatter of about 50 per cent. ; Comment: 13 pages, 11 figures, submitted to MNRAS