A review of recent research on the non-medical control of auditory hallucinations is presented. It is suggested that the decreases in hallucinatory behavior obtained in studies using aversive contingencies may be attributable to the disruption of the chains of behavior involved. The results of several additional studies are interpreted as indicating that methods of stimulus control and the use of incompatible behaviors may be effective in reducing the rate of auditory hallucinations. Research relating auditory hallucinations to subvocalizations is presented in support of the view that hallucinatory phenomena are sometimes related to the subject's own vocal productions. Skinner's views (1934, 1936, 1953, 1957, 1980) are then presented as possible explanations of some hallucinatory behavior. It is suggested that some auditory hallucinations consit of the mishearing of environmental and physiological stimuli as voices in a fashion similar to that which Skinner observed in his work with the verbal summator. The maintenance of long chains of such responses may be largely attributable to self-intraverbal influences (such as are present during automatic writing). With some auditory hallucinations, this progression involves first mishearing ambiguous stimuli as voices and then attributing the voices to some cause (e.g., insanity, the television, radio, or God). Later, the frequent and ongoing chains of such behavior may contaminate other verbal responses. Such verbal behavior may be parasitic on "normal verbal behavior" (and hence, not directly dependent on consquences for maintenance), may be cued by various stimuli (including respiration), and may interfere with other covert and overt behavior. Several studies to investigate this view are presented. It is hoped that such research will lead to a better understanding of the major issues involved in the etiology and treatment of auditory hallucinations in particular and perhaps of psychosis in general.