A correct estimation of the net infiltration is of prime importance to know the amount of water that is available for solute transport. For a bare soil surface under natural weather conditions, net infiltration is defined as the difference between precipitation and actual evaporation per unit of time. The actual evaporation does not only depend on the atmospheric water demand, i.e., the potential evaporation, but also on the ability of soils to sustain the evaporative flux, i.e., the soil hydraulic properties. Numerical simulation models solving flow problems normally require precipitation and potential evaporation rate to impose a transient upper boundary flux. Actual evaporation is calculated internally. Potential evaporation is estimated from micro-climatological data, such as temperature, net radiation, relative humidity and wind speed, using empirical or physically-based models. The parametrization of both the evaporation and hydraulic property models is site-specific and a priori unknown. An improper parametrization of both models can mutually cancel out the effect on the actual evaporation. We will illustrate this dilemma and present the results of a non-linear parameter optimization analysis (Levenberg-Marquardt) using data from a 1.2-m high, weighable, bare soil lysimeter, equipped with 10 TDR probes at five depths. The lysimeter was exposed to climatic boundary conditions for three years. The parameters of various evaporation models and a hydraulic property model were determined using the 3-years time series of the cumulative outflow, the cumulative actual evaporation, and the volumetric water content. An unequivocal determination of all parameters was only possible, when the volumetric water content data were included in the objective function. The simulations showed a weak correlation between the parameters of the evaporation and the hydraulic property model. A large part of the daily evaporation was explained by temperature; the relative humidity turned out to be insensitive.