Genistein is an isoflavone found in soy, and its chemo-preventive and -therapeutic effects have been well established from in vitro studies. Recently, however, its therapeutic actions in vivo have been questioned due to contradictory reports from animal studies, which rely on rodent models or implantation of cell lines into animals. To clarify in vivo effects of genistein in advanced prostate cancer patients, we developed a patient-derived prostate cancer xenograft model, in which a clinical prostatectomy sample was grafted into immune deficient mice. Our results showed an increased lymph node (LN) and secondary organ metastases in genistein-treated mice compared to untreated controls. Interestingly, invasive malignant cells aggregated to form islands/micrometastasis only in the secondary organs of the genistein-treated groups, not in the untreated control group. To understand the underlying mechanism for metastatic progression, we examined cell proliferation and apoptosis on paraffin-sections. Immunohistological data show that tumors of genistein-treated groups have more proliferating and fewer apoptotic cancer cells than those of the untreated group. Our immunoblotting data suggest that increased proliferation and metastasis are linked to enhanced activities of tyrosine kinases, EGFR and its downstream Src, in genistein-treated groups. Despite the chemopreventive effects proposed by earlier in vitro studies, the cancer promoting effect of genistein observed here suggests the need for careful selection of patients and safer planning of clinical trials.