Establishing a contingency between stress and a physiological response is essential in biofeedback. The sensitivity of high alpha to contingent stress was investigated by manipulating conditions known to influence stress, such as the distribution, predictability, and controllability of stressful stimuli, and number of tasks performed. Forty subjects were divided into stress and non-stress groups. Within each group, one-half had the dual-task of anticipating and increasing alpha activity. The other half was initially instructed to only anticipate alpha and, later, had the dual task of anticipating and controlling alpha. No feedback training was included to distribute the task-related stressor and allowed the assessment of self-control. All of the stress manipulations significantly influenced the effects of stress on alpha production. The dual-task subjects produced less alpha and less self-control than did training with control phased in after subjects learned to anticipate alpha. Without stress, phased-in control produced highly significant increases in alpha production and self-control without feedback. The use of an alpha-contingent feedback paradigm and anticipation training was related to the therapeutic applications of alpha feedback to stress and anxiety.