BACKGROUND. Tissue inhibitors of nietalloproteinases (TIMPs) act as negative regulators of matrix metalloproteinases (MMPs) that degrade extracellular matrix. We evaluated the metastatic ability of the highly metastatic human gastric cell line KKLS, and that of cells transfected with exogenous TIMP-I gene by the orthotopic transplantation model in nude mice. METHODS. KKLS was derived from human gastric cancer. Expression of mRNA for tissue inhibitor of nietalloproteinase-1 (TIMP-1) was almost undetectable in KKLS cells. KKLS cells were transfected with exogenous TIMP-1 gene by the Chen-Okayama method. Two clones (KTCLs) that expressed different levels of TIMP-1 and neomycin-resistant KKLS (K-neo) were obtained. The KKLS cells and these transfectants were orthotopically transplanted into nude mice (murine stomach) and metastasis in the murine liver was detected. As a method of detecting metastasis, we used a DNA fragment (human P-globin gene) specific to human tumor cells that have metastasized into the murine liver by polymerase chain reaction (PCR). RESULTS. Differences in tumor growth in the murine stomach were not observed between KKLS cells, K-neo cells, and the two transfectants expressing the different TIMP-1 levels (low, KTCL-1; high, KTCL-14). The KKLS cells and K-neo cells had undergone liver metastasis, as shown by PCR amplification of the human ,&globin gene fragment from the murine liver samples, since Week 1 after transplantation and the metastasis had grown exponentially; however, although KTCL-1 cells and KTCL-14 cells had undergone liver metastasis since Week 2, the metastasis had not grown. The average intensities of the amplified gene fragments from K-neo cells, KTCL-1 cells, and KTCL-14 cells in Week 4 after transplantation were loo%, 45%, and 18%, respectively, of the parental KKLS cells. CONCLUSIONS. TIMP-1 was suggested to act as a negative regulator of the nietastasis. The present data is thought to be especially important because the mice in this study underwent orthotopic transplantation with a metastatic model.