Many tumors target the specific immune system cells raised to protect the host against tumor. This targeting can take two main shapes. Either the tumor cells attract immune cells to the tumor, enhancing the tumor growth, or they prevent the immune cells from killing existing tumor cells. In both cases a positive feedback loop emerges between the tumor and immune system cell concentrations. Such a feedback loop may explain the equilibrium obtained between the host immune system and tumors, where tumors stop growing, or grow very slowly, but are not destroyed by the immune response. While tumor size can increase by a factor of 10 within a day, in most cases, this huge division rate is not obtained and in reality the tumor is almost in equilibrium. This equilibrium can be explained by a bi-stable solution of the tumor-immune system dynamics. We here study a generic set of feedback loops between the immune system cells and their targets in tumors and show that a bi-stable solution can emerge. This solution can occur only if the tumor induces death or inactivation of macrophages. In such a case, a simple negative effect of the pathogens on the macrophages will suffice to induce bistability. The initial conditions then becomes crucial for the solution, given that, according to it, the solution tends to one or the other fixed point, which correspond to a healthy or a sick state. We show that double inhibition positive feedback loops (immune system kills tumor, which in turn kill immune system cells) behave differently than double activation feedback loops (Tumor produce cytokines that attract immune cells, which in turn produce cytokines which induce tumor growth).Double inhibition feedback loops induce bi-stability in most parameter space, while double activation feedback loops induce such a bistability in very limited regions of parameter space.