Passive capillary lysimeters (PCLs) are uniquely suited for measuring water fluxes in variably-saturated soils. The objective of this work was to compare PCL flux measurements with simulated fluxes obtained with a calibrated unsaturated flow model. A Richards equation-based model was calibrated using TDR data from a loamy soil profile involving 12 replicates along transects at five depths.  The fluxes themselves were measured at depths of 15 and 55 cm, and were estimated at the depth of 105 cm using the soil storage dynamics. Annual cumulative fluxes were used to compare measurements and simulations. Large discrepancies were found between simulated and measured fluxes. In an attempt to explain these discrepancies, the fluxes were also computed using a simpler, multiple bucket-type, soil-water model (MWBUS) which was calibrated independently with the TDR soil water monitoring data. PCL measurements and the MWBUS-simulated results were in a very good agreement. Further analyses showed that improvements in the fit of the multiparametric Richards model was achieved by introducing water losses to runoff, whereas no runoff was observed in the experiment. The simulated loss of water to runoff apparently distorted simulated fluxes in the soil profile.  This study demonstrated that PCL flux measurements can be extremely useful for improving modeling of variably saturated flow in soils.