The soil water retention characteristic, $theta$(h), is required for modeling and predicting water and solute transport in unsaturated porous media. Commonly, $theta$(h) is determined by relating pressure heads, h, to mean water contents, $theta$, that are measured in column experiments under hydrostatic equilibrium conditions and fitting a parametric retention function to these data pairs. Implicit to this method is the assumption that the mean water content of the column is equivalent to a point measurement in the column center. Dependent on the nonlinearity of the vertical water content distribution, $theta$(z), in the column, this assumption may be invalid and introduces a systematic error. A sensitivity analysis shows that the magnitude of the error caused by neglecting the nonlinearity of the water content distribution may reach several percent if coarse materials with low air entry values and tall soil columns are investigated. Furthermore, neglecting $theta$(z) yields a smoothed retention characteristic and thus may lead to wrong conclusions about the most appropriate parametric model for the water retention characteristic. If the hydraulic conductivity function K(h) is predicted from such an incorrect retention function, it can differ greatly from the true function. In this paper, we propose to consider the measured water content of a soil column explicitly as an integral of the equilibrium water content distribution with depth. We show that this eliminates systematic parameter estimation errors and leads to improved estimates of the soil water retention function.