Abstract Limiting factors in systemic recombinant interleukin-2 (rIL2) therapy may be overcome by intratumoral (IT) administration. A series of experiments was conducted to assess the efficacy of IT rIL2 alone and in combination with LAK cells and IFN-γ. C57BL/6 mice bearing B16-F10 subcutaneous tumors were randomly assigned to treatment groups including: noninjected controls, IT placebo (NaCl, D5W), IT bovine serum albumin (BSA), IT rIL2 (centrally and peripherally), IT rIL2/LAK, IT rIL2/IFN-γ, and intraperitoneal (IP) rIL2. A tumor size-dependent dose of cytokine was injected daily and LAK cells were given weekly. Systemic immune response was assessed by splenocyte mitogenesis and T-cell subset distribution using thymidine radioassay and flow cytometry, respectively. In terms of survival and tumor growth rate, IT rIL2 was superior to noninjected control, IT placebo, IT BSA, and IP rIL2 ( P < 0.05). The addition of IT LAK cells conferred no therapeutic advantage. The combination of rIL2 and γIFN-γ had a slight survival benefit over rIL2 alone (30.8 days vs 20.4 days). Histologic analysis demonstrated an increase presence of intratumoral macrophages in the IT rIL2-treated tumors ( P < 0.05). Lymphocyte mitogenesis and L3T4 + subset were not altered by any treatment. In vitro thymidine uptake by tumor cells was not affected by rIL2 nor IFN-γ alone but the combination of rIL2 and IFN-γ resulted in significant tumor cell growth inhibition. Spontaneous lung metastases were more prevalent following central IT rIL2 (75% vs 29%, P = 0.07) not accountable by needle trauma but avoidable by the use of peritumoral injection. In conclusion, local administration of rIL2 exerted greater antitumoral effect without systemic toxicity. This effect may be mediated by local tissue macrophages unrelated to nonspecific inflammatory response. Peritumoral injection appears to be the optimal method of administration.