A stochastic model is proposed for the description of the hydrodynamics in trickle flow columns. The gas-liquid flow is assumed to be controlled by local packing properties which are represented by random variables. The first (microscopic) level of description corresponds to the smallest volume whose averaged geometrical properties are representative of the whole packing. The hydrodynamics at this level is described by Ergun equations. The second level describes the liquid flow maldistribution induced by the partial wetting of the solid. This is the macroscopic level where the hydrodynamics is described by a probability distribution. This approach is used to develop correlations for predicting the wetting efficiency, the dynamic liquid holdup, and the pressure drop. These correlations give very good results when compared to experimental data over a wide range of operating conditions. In addition, we propose to describe the hysteretic behavior of trickle beds by a modification of the loquid flow texture caused by different values of the apparent solid wettability.