Three independent laboratories tested eight "model" and five coded chemicals in the Syrian hamster embryo clonal transformation assay system to establish the intra- and interlaboratory reproducibility of the system and to identify sources of variability. When a common cell pool and the same lot of fetal calf serum were used, the three laboratories obtained consensus on the activity of eight model chemicals: five chemicals (benzo(a)pyrene, 7,12-dimethylbenz(a)anthracene, N-methyl-N'-nitro-N-nitrosoguanidine, nitroquinoline-N-oxide, and lead chromate) induced morphological transformation without exogenous metabolic activation and three (N-2-fluorenylacetamide, pyrene, and anthracene) produced no transformation response. Five coded chemicals (2,6-dichloro p-phenylenediamine, 4,4'-oxydianiline, cinnamyl anthranilate, dichlorvos, and reserpine), representative of environmental chemical classes, but not necessarily strong carcinogens, produced more equivocal responses in this interlaboratory study. Thus, while the assay can be used to distinguish between transforming and nontransforming chemicals in some cases, the intrinsic limitations in low transformation frequency and in achieving any dose-response results are major constraints to the use of this system in a routine testing program at the present time. Efforts to increase the transformation frequency or to amplify the expression of the transformed phenotype constitute some of the approaches which should be explored in order to overcome these limitations.