Design of experiments (DOE) is a value scientific approach used to understand the processes in a better way and to determine how the inputs affect the response(s). In this work, this method has been applied to study the behavior of the three-phase distillation in a sieve tray column, through the effects of the process and geometrical variables. The experimental values were compared with predicted values obtained by simulation using the equilibrium and nonequilibrium models. Three-phase distillation has been used for glycerine dehydration using toluene as entrainer, in order to avoid the glycerol degradation by distillation at atmospheric pressure. The best conditions found were: vapor flow rate of toluene=23.5kg/h, feed flow rate=2.2kg/h and feed concentration=50wt% glycerol, using the layout L4 with fractional hole area=0.04 and weir height=70mm. The nonequilibrium model based on Eckert and Vanek's approach (2001) and Chen-Chuang's correlation (1993) have been used to estimate the binary coefficients of mass transport. The predicted values obtained by the experimental model and by the nonequilibrium model have represented the behavior of the dehydration in the sieve tray column studied. Both models underpinned the experimental results obtained for this column.