Fifty patients with primary localized lung cancer were tested at the time of surgery for the ability of their lymphocytes to kill autologous, freshly isolated tumor cells, and the assay was evaluated for prognostic significance. Peripheral blood lymphocytes of 27 patients (54%) demonstrated significant autologous tumor-killing activity in 6-hour 51Cr-release assays. Twenty-three of the 27 patients with autologous tumor-killing activity remained tumor free and survived more than 5 years after curative surgery, while all 23 who were negative for autologous tumor-killing activity relapsed by 18 months after surgery and died within 42 months after surgery. The differences in survival curves for the two groups were highly significant (P less than .00003). Autologous tumor-killing activity was not correlated with natural killer (NK) cell activity against K562 human myeloid leukemia cells or proliferation of lymphocytes stimulated with autologous, freshly isolated tumor cells in mixed culture. There were no differences in total survival between patients with positive results and those with negative results in tests of NK cell activity and autologous mixed lymphocyte-tumor culture reaction. These results indicate that autologous tumor-killing activity is a meaningful prognostic indicator and provide evidence for immunological control of tumor growth and metastasis. According to our preliminary data, it is unlikely that lung cancer patients who remain tumor free after 60 months of follow-up will develop recurrence or die from the disease. We are conducting a study to determine whether induction of autologous tumor-killing activity before surgery, by treatment with biological response modifiers,can improve the clinical outcome in patients who do not naturally have this potential.