This study confirms our earlier finding that human interleukin (IL)-1 beta exerts an antiviral effect on diploid fibroblasts and on MG-63 osteosarcoma cells. It also extends the observation in that a similar effect was noted on aged but not freshly trypsinized HEp-2 cells, and that not only IL-1 beta but also IL-1 alpha and tumor necrosis factor (TNF)-alpha exerted similar antiviral effects on cells. The antiviral effects of these cytokines were neutralized by addition to the assay system of an antibody that was specific for interferon (IFN)-beta 1, indicating that IFN-beta 1 or a structurally or functionally related substance is involved in the antiviral activity observed. Both IL-1 and TNF were able to induce production of the 26-kDa protein, also known as IFN-beta 2, hybridoma/plasmacytoma growth factor (HPGF) or B-cell stimulatory factor-2 (BSF-2) and previously proposed as an alternative to IFN-beta 1 for mediating the antiviral effect of TNF. However, no good correlation was found between the antiviral effects of TNF and its potential to induce production of the 26-kDa protein. Furthermore, the anti-IFN-beta 1 serum which neutralized the antiviral activity of IL-1 and TNF did not cross-react with the 26-kDa protein. Conversely, the antiviral effect of IL-1 and TNF was only weakly neutralized by an antibody that did react with the 26-kDa protein and showed low cross-reactivity with IFN-beta 1. These observations, together with the low specific activity of the 26-kDa protein as an antiviral agent (less than 10(5) U/mg protein) provide strong arguments against this protein and in favor of IFN-beta 1 (or still another IFN-beta 1-related molecule) as the ultimate mediator of the antiviral effect of IL-1 and TNF.