Activation of peripheral T cells by foreign and self-antigens is under stringent control by different mechanisms, both thymic and peripheral, and their failure may lead to autoimmune diseases. The progress in understanding the mechanisms of T cell activation and inactivation is currently being translated into strategies able to induce selective immunosuppression to treat different pathological situations, notably autoimmune diseases, as well as allergies and allograft rejection. The medical need for selective immunosuppression is very high, as the available immunosuppressive drugs are substantially inadequate because of limited efficacy, modest selectivity and considerable toxicity. Key attack points for selective immunointervention have been identified: modulation of antigen recognition, co-stimulation blockade, induction of regulatory cells, deviation to non-pathogenic or protective responses, neutralization of proinflammatory cytokines, induction or administration of anti-inflammatory cytokines and modulation of leucocyte trafficking (Table 1). Therefore, to interfere selectively with the activation of pathogenic T cells, immunosuppressive therapy can be directed to three cellular targets: antigen-presenting cells, autoreactive T cells and suppressor/regulatory T cells, with the common goal to selectively inhibit the activation of pathogenic class II-restricted CD4+ T cells. Table 1 Selective immunointervention in autoimmune diseases All these forms of immunointervention have been used successfully to prevent and sometimes treat experimental autoimmune diseases. Based on these results, expectations have been raised for exploiting the same strategies to inhibit the activation of human autoreactive T cells. In this review, we will examine recent advances towards cytokine-based immunointervention in autoimmune diseases, highlighting their successes and failures in clinical applications. Indirect approaches to manipulate the cytokine network that rely on protocols based on oral tolerance or administration of altered peptide ligands to modify the pattern of cytokine production will not be discussed because of space constraints, but they have been reviewed recently [1–3].