The purpose of tumorigenicity testing, as applied not only to cell substrates used for viral vaccine manufacture but also stem cells used for cell-based therapy, is to discriminate between cells that have the capacity to form tumors and cells that do not. Therefore, tumorigenicity testing is essential in assessing the safety of these biological materials. Recently developed NOD/Shi-scid IL2Rg^null (NOG) mice have been shown to be superior to NOD/Shi-scid (SCID) mice for xenotransplantation of both normal and cancerous cells. To select a suitable mouse strain as a xenogenic host for tumorigenicity testing, we compared the susceptibility of NOG (T, B, and NK cell-defective), SCID (T and B cell-defective), and the traditionally used nude (T cell-defective) mice to tumor formation from xenotransplanted HeLa S3 cells. When 10⁴ HeLa S3 cells were subcutaneously inoculated into the flanks of these mice, the tumor incidence on day 22 was 10/10 (100%) in NOG, 2/10 (20%) in SCID, and 0/10 (0%) in nude mice. The subcutaneous tumors formed reproducibly and semiquantitatively in a dose-dependent manner. Unexpectedly, half of the NOG mice (5/10) that had been inoculated with a mere 10¹ HeLa S3 cells formed progressively growing subcutaneous tumors on day 78. We confirmed that the engrafted tumors originated from inoculated HeLa S3 cells by immunohistochemical staining with anti-HLA antibodies. These data suggest that NOG mice may be the best choice as a suitable strain for testing tumorigenicity.