Statistical techniques for the treatment and analysis of experimental data from laboratory reactors were considered, as applied to a CREC Riser Simulator reactor for the example of fluid catalytic cracking catalyst evaluation. Deviations in mass balances were reconciliated considering the variance in each product mass. After reconciliations, simple optimum performance envelope curves were used to fit the data due to their simplicity and proper representation of the yield curves. In the data fitting step, errors in both dependent and independent variables were considered by using information obtained in the reconciliation procedure. The impact of different levels of confidence bands in the models on the discrimination of experimental results was discussed. Significant improvements in catalyst evaluation could be achieved either in the CREC Riser Simulator or other types of laboratory reactors with the help of the statistical procedures described here without increasing substantially the number of experiments.