A major problem in designing adequate feedback controllers for physiological systems is in the large variability in the dynamics of individuals. Adaptive control schemes are considered to be suitable candidates to solve this clinical problem and, in this work, the control of muscle relaxation using a self-tuning PID control technique is investigated. The performance of this control strategy is evaluated in a real-time simulation environment, using published pharmacokinetics and pharmacodynamics of the recently introduced competitive neuromuscular blocking agent Atracurium. The sensitivity of the control algorithm with respect to simulated intra- and interindividual variability in the dynamics is studied and the feasibility of combining bolus dose and closed-loop infusion protocols of Atracurium administration is demonstrated. A series of clinical trials were undertaken where the self-tuning PID algorithm controlled successfully the delivery of Atracurium to patients undergoing surgery.