The identification of vadose zone flow parameters and solute travel time from the surface to the water table are key issues for the assessment of groundwater vulnerability. In this paper we use the results of time-lapse monitoring of the vadose zone in a UK consolidated sandstone aquifer using cross-hole zero-offset radar to assess and calibrate models of water flow in the vadose zone. The site under investigation is characterized by a layered structure, with permeable medium sandstone intercalated by finer, less permeable, laminated sandstone. Information on this structure is available from borehole geophysical (gamma-ray) logs. Monthly cross-hole radar monitoring was performed from August 1999 to February 2001, and shows small changes of moisture content over time and fairly large spatial variability with depth. One-dimensional Richards’ equation modeling of the infiltration process was performed under spatially heterogeneous, steady state conditions. Both layer structure and Richards’ equation parameters were simulated using a nested Monte Carlo approach, constrained via geostatistical analysis on the gamma-ray logs and on a priori information regarding the possible range of hydraulic parameters. The results of the Monte Carlo analysis show that, in order to match the radar-derived moisture content profiles, it is necessary to take into account the vertical scale of measurements, with an averaging window size of the order of the antenna length and the Fresnel zone width. Flow parameters cannot be uniquely identified, showing that the system is over parameterized with respect to the information content of the (nearly stationary) radar profiles. Estimates of travel time of water across the vadose zone are derived from the simulation results.