A 54 Kd apoptosis-inducing protein with novel amino acid sequence has been purified from the conditioned medium of the embryonic cell line, C3H 10T1/2 cells. An apoptosis-inducing protein identified to be fetal fetuin and a 60 Kd apoptosis-inducing protein have also been found in fetal serum and fresh embryo extract, respectively. Interestingly, a common characteristic of these embryonic apoptosis-inducing proteins is that they selectively induced apoptosis in cancer without affecting normal cells. For example, the 54 Kd protein selectively induced apoptosis in 10 out of 12 cancer cell lines without affecting 12 normal cell lines we tested. Fetal fetuin, on the other hand, selectively induced apoptosis in 5 cancer cell lines without affecting the 3 normal cell lines we tested. In vivo, tumor animal model study showed that fetal fetuin enhanced survival in leukemia-bearing mice and strongly inhibited the formation of prostate cancer in a PC-3 prostate cancer model in mice. A working hypothesis has been proposed to aid in the study of the mechanism by which the embryonic apoptosis-inducing proteins selectively induced apoptosis in cancer without affecting normal cells. This hypothesis states that due to the retro-differentiational characteristic of malignancy, cancer cells may re-express the signal transduction machinery for development-related apoptosis, which is otherwise to be normally expressed by embryonic, but not by adult cells. The embryonic apoptosis-inducing proteins may therefore induced apoptosis in cancer but not in normal cells and may be developed as an anticancer agent. This new concept may constitute a new approach for cancer therapy, which we tentatively designated as "Retro-differentiational Apoptosis Cancer Therapy", (R-ACT).