The objectives of the present study were to determine the safety of locoregional administration of autologous lymphocytes stimulated with autologous tumor cells and interleukin (IL) 2 in vitro and to find laboratory markers to predict either clinical toxicity or clinical response. Eleven patients with advanced (n = 4) or recurrent (n = 7) esophageal cancers received the locoregional administration of these activated lymphocytes every 2 weeks for two to nine times (mean, 5.6 times), and mean numbers of the administered cells were 0.8 x 10(9) cells per treatment. The activated lymphocytes that were pretested for their surface markers and CTL activity were endoscopically injected into primary tumor sites (n = 4) or directly injected into metastatic lymph nodes (n = 2), pleural (n = 4) or ascitic (n = 1) regions. Grade 3 hypotension, grade 2 diarrhea, and grade 1 fever were observed in 1, 1, and 6 patients, respectively, and there was no adverse effect in the remaining three patients. The clinical outcome was as follows: one, complete response (CR); three, partial response (PR); two, stable response (SR); and five, progressive disease (PD). CTL activity in the administered cells was observed in 5 of the 11 patients (1 CR, 3 PR, and 1 PD) and was not observed in the remaining 6 patients (2 SR and 4 PD). Percentages of CD16+ cells in the peripheral blood of the responder group (CR+PR) significantly increased when compared with those before treatment or with those of the nonresponder group before as well as after treatment. Because the clinical toxicity was moderate and tolerable, this new method of locoregional immunotherapy will be applicable for use in treatment of patients with advanced and recurrent esophageal cancers. Both CTL activity in the administered cells and the percentages of CD16+ cells in the peripheral blood may be useful laboratory markers for predicting of clinical response.