biofeedback is the use of instrumentation to provide feedback to a client about his/her psychophysiological processes. one type of biofeedback is electromyographic (emg) which measures muscular contraction and relaxation. this thesis examines the use of emg feedback as a therapeutic tool in teaching relaxation to seven mentally retarded individuals from a specialized treatment area for aggressive and self-injurious clients in a provincially-operated institution. experiment 1 examined the feasibility of emg biofeedback relaxation training with a 21 year old profoundly retarded male and a 27 year old severely retarded female. an auto clinic 2001 (colbourn instruments) biofeedback apparatus was used to monitor emg activity. subjects participated in 10 training sessions for each of the following three conditions: (a) contingent biofeedback (music) relaxation training; (b) control on (continuous music); and (c) control off (no music). the results demonstrated more consistent reductions of emg activity across contingent conditions for both subjects. contingent biofeedback training also resulted in significantly lower measures for one subject in comparison to one control condition, and two control conditions for the other subject. experiment 2 examined the possibility of generalizing the effects of biofeedback training across settings and compared the effects of a proportional-analog stimulus and discrete stimulus. the three subjects were a 21 year old mildly retarded male, a 31 year old severely retarded male and a 15 year old mildly retarded male. the same equipment was used, with modifications that allowed the production of a proportional-analog stimulus. all subjects were exposed to 25 sessions consisting of the following four conditions: discrete, variable, baseline and control. music was presented to the clients in the discrete and variable conditions. the results from the final sessions demonstrated that the discrete condition induced more relaxation in all subjects than the three other conditions. an analysis of the pre- and post-training rating scales indicated that 100% of the improvement in the subjects behaviour occurred in the non-control treatment conditions. experiment 3 investigated the possible advantage of massed practice emg biofeedback training with a 26 year old mildly retarded male and a 42 year old moderately retarded male. subjects were exposed to the same conditions as in experiment 2 for a total of 26 sessions. analyses of the one subject's data suggested that both feedback (treatment) conditions induced more relaxation than the baseline condition, but not significantly more than the control condition. analyses of the other subject's data suggested that the subject only relaxed during the control condition. in summary, the results of these experiments demonstrated that emg biofeedback is feasible with mentally retarded clients, and that a descrete condition was the best relaxation technique (experiment 2). it was also shown that generalization of biofeedback training across settings was not detected, and that massed practice may be more conducive to learning to relax, as opposed to randomizing and inter-mixing several conditions. several suggestions for future research were also discussed.