Sorption isotherms are essential for describing transport of solutes in groundwater, and may be obtained in column experiments. In this paper a novel algorithm for the numerical computation of the sorption isotherm from the measured data is presented. The algorithm, which is quite different from classical minimization algorithms like the Marquardt process, uses the relative arrival times of dissolved components with a different sorption behavior. It has a large domain of attraction, and it is also rather inexpensive computationally. It is most effective as a preprocessor for a classical minimization process. The applicability of the algorithm is demonstrated on elution curves of K+, displaced by Ca++ in a cation exchange reaction. The speed of the algorithm allows for the use of the bootstrap method for obtaining confidence domains for different sorption sites with different exchange constants. ¿ American Geophysical Union 1995