Interleukin-2 (IL-2) has shown antitumor activity in some neoplasms, such as melanoma and renal carcinoma, but toxicity derived from bolus administration is significant, particularly at the cardiorespiratory level. To test feasibility, antitumor activity, pulmonary and systemic immunologic effects, and pulmonary function changes of continuous-infusion recombinant IL-2 given to patients with non-small cell lung cancer, eleven subjects with Stage III-IV disease were treated in a standard pulmonary medicine unit with a dose of 18 million IU/m2/day from day 1 to day 13 with 1-day rest on day 7. A second induction course was given after a 3-week rest. In patients with nonprogressive disease, four maintenance courses of 6 days' duration at the same dose were planned. Immunologic tests, including lymphocyte phenotype analysis and assays for the detection of tumor necrosis factor (TNF) and of anti-IL-2 antibodies, were performed before and after treatment in serum and bronchoalveolar lavage fluid (BAL). Cardiopulmonary function tests, including spirometry, arterial blood gas analysis, diffusion capacity, and echocardiography, were obtained before, during, and after treatment. Twenty-one cycles (15 induction courses plus 6 maintenance courses) were administered. No patient was able to complete the six planned courses, and only 3 patients entered the maintenance phase. Reasons for discontinuation included progressive disease in five cases, toxicity in three cases, and patient request in three cases. The most common side effects were fever, hypotension, oliguria, and elevated serum creatinine and liver enzyme levels. No patient required intubation or intensive care. No objective response was seen, and the median survival time was 10 months. Lymphocytosis and eosinophilia were observed in all patients. Surface marker analysis revealed a statistically significant increase in the percentage of CD3+, CD4+, CD25+ and DR+ cells in peripheral blood. Lymphoid cells derived from BAL disclosed an increased natural killer activity after IL-2 treatment, and TNF was increased in BAL fluid. Pulmonary function tests evidenced an increased alveolar-arterial difference for oxygen allied with a decrease of forced expiratory volume in 1 second, forced vital capacity, and carbon monoxide transfer coefficient consistent with a significant, albeit not clinically relevant, interstitial lung defect. Continuous-infusion IL-2 is feasible in patients with advanced lung cancer even outside an intensive care unit, but overall compliance is poor. Although clinical pulmonary toxicity is negligible, small but statistically significant alterations of the pulmonary function are evident. In addition, this regimen produces a significant activation of the immune system at the pulmonary level.