Immunotherapy refers to the use of natural and synthetic substances to stimulate the immune response. This paper provides a description for an adaptive control design for immunotherapy cancer treatment mathematical model, where all state vector is considered to be not on-line available. The control strategy is suggested in two parts: the first one deals with the state estimation process using differential neural networks and sliding mode type observer techniques. The second part introduces the non-linear neural-identifier design which provides a complete state information to construct a discontinuous (sliding mode type) control function. This technique was successfully applied in the tracking process for immunotherapy dosage control.