We had previously shown that the signal of activation delivered via CD2 varies according to the mitogenic pair of CD2 mAb used. We had selected two typical mAb pairs, D66 + T11(1) and GT2 + T11(1), the former delivering the "richest" signal, the latter the poorest. Here we analyzed the cytolytic activities generated within PBL-stimulated by these two pairs. When purified CD2+,3- cells were cultured with either one of these two pairs, no significant lymphokine-activated killer (LAK) activity--namely the activity exerted on NK-resistant malignant cell lines or fresh tumor cells--was detected, thereby demonstrating the inability of CD2 mAb pairs to directly trigger the LAK precursors. By contrast, when purified CD2+,3+ cells were cultured, only D66 + T11(1) was able to trigger a potent CTL activity, as judged by targeting their activity, at the effector phase, with a bridging CD3 mAb on a FcR+ target cell or by using heteroaggregates on FcR- malignant cells. When whole PBL were used, a similar and moderate LAK activity was generated after culture with either one of the 2 CD2 mAb pairs. This, in fact, masked quite different events. The amounts of endogeneous IL-2 released in PBL cultures with GT2 + T11(1) was rather low, although it was sufficiently high in PBL cultures with D66 + T11(1) to generate a potent LAK activity. Yet, PBL stimulated with D66 + T11(1) released concomitantly a high amount of IL-4 which inhibited the development of the LAK activity, as demonstrated by unmasking this activity with an anti-IL4 antiserum and which did not inhibit the T CTL activity; this IL-4 secretion was not seen with GT2 + T11(1). Therefore, stimulation by these two typical CD2 mAb pairs induce a striking different pattern of IL synthesis.