Abstract: The current research has studied the effectiveness of relaxation training conducted via imagery. Training with preferred and nonpreferred sensory modalities have been compared on the basis of both subjective and physiological measures. Imagery is a psychological phenomenon which has a long history in western thought but a limited history in physiological research. Imagery in behavioral psychotherapeutic methodology has traditionally been thought of as existing in only one sensory modality: vision. George Betts (1909) formulated a vividness of imagery scale (QMI), and Sheehan (1967) shortened Betts' original instrument and established its reliability. The hypnotherapist Milton Erickson believed that the sensory system which an individual emphasizes in representing his/her world is an important factor both in building rapport and in eliciting therapeutic change. Richard Bandler, John Grinder, and others have articulated and promoted a set of techniques, neurolinguistic programming (NLP), in which the concept of primary representational system -- or preferred sensory modality --is crucial. Each of 134 subjects in the current research participated in one forty-minute session. Betts' QMI yielded scores indicating the vividness of each subject's imagery in three modalities: visual, auditory, and tactile/kinesthetic. Subjects first relaxed on their own, and then received treatment consisting of beach imagery described in explicit terms representing one sense mode. Subjective and physiological measures were taken both before and after treatment. The subjective measure was Spielberger's (1969) State Anxiety Scale, and the physiological measure was frontalis muscle EMG. The data were analyzed so as to answer the question of whether preferred modality treatment groups differed significantly from nonpreferred modality treatment groups. A linear models approach was used to test the hypotheses. On the subjective measure, preferred and nonpreferred modality treatment groups both showed a decrease in anxiety, with no difference between groups. On the physiological measure, there was a significant difference between groups, such that treatment with preferred modality was more effective. The results were interpreted as indicating that use of preferred sensory modality in treatment improves relaxation training. Tentative support is thus provided to the theory suggested by Erickson and articulated by Bandler and Grinder. Further research is warranted on the basis of this study.