Cytokine-induced killer (CIK) cells were examined for safety and efficacy for cholangiocarcinoma treatment. Several conditions of human CIK cells were examined using ex vivo cytotoxic assay and SCID mice pre-inoculated with cholangiocarcinoma cells. We monitored the ex vivo cytotoxicity, tumor sizes and immunohistochemistry. Optimal tumor suppression was observed when CIK cells were pre-exposed to dendritic cells (DCs). Unexpectedly, pulsing of tumor RNA to DCs rendered the co-culturing CIK cells ineffective and raised the proportion of CD4(+)CD25(+) subset. The use of CD3(+)CD56(+) subset instead of the whole population of CIK cells for the co-culture with RNA-pulsed DCs restored the efficacy. Tumor-infiltrating human CD3(+) cells were observed from day 2 - 14. The CD3(+)CD56(+) cells are logical candidates for clinical trial while the DC-co-cultured CIK cells produced similar efficacy and more feasible for clinical application. The RNA pulsation of DCs up-regulated the regulatory subset of CIK cells and abrogated the anti-tumor efficacy.