A procedure for analysis of experimental data and model discrimination in regression of rate data of reversible, heterogeneous reactions which uses the power-law rate expression is described. The numerical values of the power-law parameters are calculated first by regression of the experimental data. These parameters are used to assess whether the quality of the experimental data is sufficient to render the following mechanistic model discrimination process conclusive. The parameter values of the power-law rate expression are also used for initial screening of the mechanism-based rate expressions, to discard infeasible models. The use of the proposed procedure is demonstrated using two examples from the literature.