SummaryTwenty-nine patients receiving recombinant interferon (IFN-α2; Schering Plough Corp., Bloomfield, NJ) were studied for changes in natural killer (NK) activity measured by a 4-h51Cr release assay against K562 cells, and T-cell subsets were determined by indirect immunofluorescence of Leu series monoclonal antibodies (Becton-Dickinson, Mountain View, CA). Seventeen cancer patients received daily i.m. injections of IFN-α2from 3 to 100 X 106U/day for 28 consecutive days or to tolerance. Twelve of an anticipated 16 melanoma patients have been studied during a phase I trial using the i.v. route with the same recombinant IFN-α2. NK activity rose during the first week of i.m. therapy from 49 ± 6.5 to 67 ± 6.2 (mean ± SE, day 8) at both high (≥30 X 106U/day) and low (≤10 X 106U/day) doses. This trend was not observed during therapy by the i.v. route at similar doses, in which NK activity tended to decrease in patients receiving 30 X 106U/day or more. Changes in T-cell subsets were observed in both trials; Leu 3a/2a (helper phenotype/suppressor phenotype) ratio rose twofold in patients receiving i.m. IFN-α2at higher doses. A rise in Leu 3a+and a fall in Leu 2a+T cells account for the change. No change in T-cell subsets was seen in patients treated at low doses (≤107U/day) by the i.m. route. By contrast, the Leu 3a/2a ratio fell by 50% in patients who received 30 X 106U/day or more of IFN-a, by the i.v. route, reflecting a fall in Leu 3a+cells and a rise in Leu 2a+cells. Thus, opposite changes in several parameters of immune competence occurred during treatment of patients with melanoma and other cancers, with a single recombinant IFN subspecies given by two different routes. These findings suggest that immunological monitoring may be critical to the optimal use of interferon.