A data-based approach to formulating quality-control criteria for the mouse lymphoma cell forward mutation assay is described. Quality-control guidelines for solvent controls, positive controls, and compound-treated cultures were developed based on analysis of over 800 experiments. Frequency distributions of experimental parameters of control cultures, such as mutant frequencies, cloning efficiencies, and suspension growths, were examined. Cloning efficiency and relative total growth affected the variability only when the test chemical was highly toxic. This information was used to generate the quality-control criteria, which were applied to an experiment before it was evaluated for a response. The response categories for classifying the effect of test chemicals on the assay system are defined in terms of (1) the statistically significant differences in average mutant frequency between solvent control cultures and cultures exposed to a chemical and (2) the trend of the dose-related responses.