The main roles which are the concern of a clinical anaesthetist are those of drugs induced unconsciousness, muscle relaxation, and analgesia. The first two roles are concentrated in the operating theatre, while the third role is mainly concerned with postoperative conditions. Unlike, measurement of unconsciousness and analgesia the measurement of muscle relaxation process is considerable easier via monitoring of evoked electromyogram (EMG) signals. Among the features characterizing this process, time delay in initiation of muscle relaxation is perhaps the most challenging one. This time delay resulted from the drug circulation around the body and variation of the cardiac output. Another problem called nonlinearity mismatch that is resulted from the wide variability of identified models and their nonlinearity in the socalled pharmacodynamics for relaxant drugs behavior. This nonlinearity is due to the large inter-individual and intra-individual variability of the patient's parameters. These challenges can be treated with quantitative or qualitative techniques. The former was proved ineffective in trying to overcome these challenges. This paper proposes predictive self-organizing Auto Regressive eXogenous (PSO-ARX) scheme to deal with such challenges with ease. This is due to two notable features of the proposed scheme one is its plastic structure and the other is its small computation required compared with Generalized Predictive Control (GPC) schemes. Simulation results reflect the superiority of the proposed PSO-ARX scheme with respect to such schemes.