Previously the sorption of riboflavin by Amberlite FPX66 resin from lactose model solutions was studied in batch system. The solutions of the transient diffusion equation were presented with a constant diffusivity D, independent of riboflavin concentration in the solution (C) or in the resin (q). The solution approximation and the constant D assumption were valid only for a fractional uptake less than 30% and greater than 70%. In order to interpret correctly the full sorption curve, in this work the equilibrium isotherm was used to express the concentration dependence of the diffusivity. The equilibrium isotherm of riboflavin on the resin at 30 degree was found to obey the Langmuir model {(q/qs)= [0.298C/(1 + 0.298C)] } and the Freundlich model {q = 6.012Cexp(0.733)}. The diffusion equation with Langmuir isotherm was solved numerically using the Crank-Nicholson method and compared with the experimental results to determine the optimal surface diffusivity, to minimise the sum of the squared deviation of predicted and measured C values. It was found that the diffusivities of riboflavin in the decalcified whey permeate on the resin were within the range of (exp-12) and (exp-11) m2/s and two times smaller than those of riboflavin on activated carbon, which agreed well with the plots of the uptake curve plot of two different adsorbent resins. The comparison suggested that Amberlite FPX66 could replace activated carbon with a minor loss in performance. As a trade off the resin would reduce the health hazard caused by black dust released from the activated carbon.