Lymphocytes infiltrating human ovarian carcinoma obtained directly from the tumour mass (tumour-infiltrating lymphocytes, TIL) or from the carcinomatous ascites (tumour-associated lymphocytes, TAL) were expanded in vitro in long-term cultures with interleukin-2 and tested for their specific cytolytic activity. Killing of the autologous tumour was detected only in a proportion of the patients, less frequently in TIL compared to TAL. In fact two out of ten TIL and four out of nine TAL cultures tested showed significant levels of lysis against the autologous tumour. This cytotoxic activity was not restricted to the autologous tumour, as other tumour cell lines, including non-ovarian ones, were lysed as well. The cultures that were not cytotoxic against the autologous tumour were in most cases able to lyse other tumour cell lines of ovarian or other histology. Cloning of TIL from one patient was performed: of 22 clones tested, 4 displayed higher cytotoxicity against the autologous tumour compared to the uncloned population and 3 out of these 4 did not kill an irrelevant carcinoma cell line. In order to stimulate the expansion of putative specific effectors we performed mixed lymphocyte/tumour cultures (MLTC) with autologous or allogeneic tumour cells. No stimulation of cytotoxicity against the autologous tumour was detected after MLTC in nine different TAL populations, using autologous or allogeneic tumours as stimulators. On the contrary, peripheral blood lymphocytes from two patients after MLTC with the autologous tumour showed increased killing of the autologous and decreased killing of an allogeneic target. In conclusion TIL and TAL from ovarian carcinoma expanded in vitro with interleukin-2 usually have non-MHC-restricted cytotoxicity and variable degrees of reactivity against the autologous tumour. A preferential killing for the autologous tumour was not observed even after MLTC. These results do not exclude the existence of tumour-specific cytotoxic T lymphocytes in ovarian carcinoma; nevertheless they suggest that putative specific effectors have very low frequency and that culture techniques for expanding their growth more selectively are still to be optimized.