On the basis of Lang's (1979) theory of emotional imagery three experiments were conducted to investigate the relationship between slow cortical potentials (SP) and emotional imagery. According to the assumptions of Lang's theory and our model (Rockstroh et al., 1982) of SP-function imagery ability should be related to a person's capacity to generate and suppress preparatory activity in cortical networks "on demand." In order to test this hypothesis subjects in Experiment I were trained to regulate right- versus left-hemispheric SP-differentiation within an instrumental learning paradigm. Thirty-four subjects were reinforced for achieving maximal SP-differences between electrode locations C3-C4 over a 6 s interval across 120 trials. Success in the SP-regulation task correlated significantly (r = .37) with the capacity for vivid imagery as measured with the Questionnaire for Mental Imagery (QMI). In Experiment II instructions to imagine right- versus left-hand movements were introduced successively over 5 sessions of SP-self-regulation. Imagery clearly modified right- versus left-hand EMG-differentiation but had no influence on cortical SP-differentiation. Experiment III tested the influence of already achieved SP-regulation at the vertex on the perceived vividness of emotional images introduced after the SP-biofeedback training. Again, clear effects of imagery content on autonomic variables (HR, SCR) were found. However, SP-amplitude and SP-polarity had no effect on perceived vividness, arousal or emotional content. It may be concluded from the results of Experiment II and III that SPs either are not the crucial parameter to represent the cortical efferent outflow component of imagery, or that the dual task of SP-self-regulation and imagery prevented covariations to show up. Experiment I, on the other hand, points toward a positive relation of imagery-ability as a trait-variable and brain-self-regulation abilities.