In Phase One, 44 subjects participated in a 2 (instructions to increase alpha, no instructions to increase alpha) × 2 (alpha biofeedback, no alpha biofeedback) factorial experiment. Results indicated that increases in alpha production were due to instructions to increase alpha and that biofeedback had no effect on alpha production. In Phase Two, the 44 subjects from Phase One were exposed to a threat of shock whereas 11 additional subjects in a control condition were not. The design employed in Phase Two was a 2 (previous instructions and stress, no previous instructions and stress) × 2 (previous biofeedback and stress, no previous biofeedback and stress) plus 1 (no previous instructions/no previous biofeedback, no stress). Results indicated that the threat of shock was effective in increasing arousal (as measured by heart rate and skin resistance) but previous EEG-alpha biofeedback training was not effective in helping subjects decrease arousal while in the stressful situation. The results indicate that it is the instructions (and related information concerning alpha) rather than the biofeedback that is critical in alpha biofeedback training and that this training does not appear to have utility for controlling arousal under stress.