We previously reported the successful treatment of the BCL1 lymphoma in Balb/c mice using bispecific (anti-CD3 x anti-idiotype) antibodies (BsAb). The in vivo effect was dependent on the bridging of tumor cells and CD3-positive cells. This direct CD3/TCR cross-linking induced targeted cytolytic and cytotoxic activity toward the tumor cells. Definitive proof of an underlying T cell-mediated mechanism was obtained, as the therapeutic effect was completely lost when animals were T cell depleted before treatment. In all these experiments, animals were injected intraperitoneally (i.p.) with 5000 tumor cells (day 0) and treated with one single intravenous (i.v.) injection of 5 micrograms BsAb (day 9). At a higher tumor load, the therapy lost its effectiveness. We evaluated repeated injections of bispecific antibodies (BsAb) to treat mice with a higher tumor burden (10(5) tumor cells). However, a dose-related immunosuppression (even with 5 micrograms BsAb) was induced. Therefore, this approach was only beneficial if the immune system could recover from the previous injection of BsAb. To prevent this induction of anergy, costimulation with bivalent anti-CD28 has been proposed, but despite the high in vitro T cell proliferation using BsAb + anti-CD28 versus BsAb alone, we were repeatedly unsuccessful in improving our in vivo results using immunologically naive animals. Only when T cell preactivation was induced was a significantly better outcome observed when the animals were treated with a mixture of BsAb + anti-CD28 compared with BsAb or anti-CD28 alone. The potency of the BsAb + anti-CD28 combination was demonstrated by the fact that 10 times less BsAb was necessary to cure animals with a 20 times higher tumor burden.