This study investigated the ability of pretreatment variables from three different domains (social-demographic, psychological, and psychophysiological) to predict posttreatment mean arterial pressure (MAP) for 59 unmedicated males with mild hypertension who were participating in a cross-cultural (USA-USSR) comparison of autogenic training and thermal biofeedback to a self-relaxation control. The overall multiple regression equation consisted of two variables and indicated that higher diastolic blood pressures during a cold pressor task were predictive of greater MAP reductions while higher scores on the Irritability subscale of the Buss-Durkee Hostility Scale were predictive of less MAP reductions. Suggestions for future research in this area are provided.