Despite the fact that most researchers acknowledge the high prevalence of comorbid substance abuse among schizophrenic patients, there is no common agreement regarding the etiology of this serious public health problem. At the center of this debate though, Khantzian's self-medication hypothesis has captured most of the attention. In the present literature review, the authors evaluate this hypothesis in the light of our current knowledge. Formulated in a clinical context, in reaction to the psychoanalytic interpretation of addiction as a pleasure seeking pathology, Khantzian's hypothesis holds that schizophrenic patients use psychoactive substances to relieve their symptoms. Properly understood, this conjecture presupposes that, with the relief of certain target symptoms, substance use would no more be a necessity. But in reality, the use of psychoactive substances usually leads to a general deterioration of the patients' condition. Pharmacodependent schizophrenic patients relapse more often, they are more frequently hospitalized, they show more violent behaviors, and they are more frequently homeless. In particular, the positive symptoms of these patients are generally exacerbated by the psychoactive drugs--with the possible exception of opiates. This observation is in lign with the fact that psychostimulants (cocaine, amphetamines), anesthesic dissociatives (PCP, ketamine) as well as hallucinogens (cannabis, LSD) are all known to exert psychotomimetic effects. As for negative symptoms, the reality is more complex. Preliminary results certainly suggest that stimulants (minor or major) relieve these symptoms, but in the case of the other psychoactive substances, empirical evidence remains fragmentary. Still, the properties of psychoactive substances invite to pay close attention, among the negative symptoms, to the cognitive deficits, the social inaptitudes and the hedonic deficits of these patients. Unsatisfied with the self-medication hypothesis, an increasing number of researchers hypothesize that schizophrenic patients abuse drugs in hope to relieve the negative affects (stress, depression) that commonly accompany their symptomatology. Interestingly, increasing data link these negative manifestations and substance abuse among schizophrenic patients. But these same data do not elucidate whether these manifestations are primary or secondary to drug abuse. For the moment, these findings must be replicated. Furthermore, it remains to be clarified what negative affect is involved here. Is it stress, anxiety or, as commonly thought, depression? Other paths aim in the direction of personality traits and dissociation. The first path is suggested by recent studies demonstrating that pharmacodependent schizophrenic patients differ from non-abusing schizophrenics in that their personality is characterized by traits such as sensation seeking and impulsivity. As for the second path, it is suggested by a recurrent observation in addictive medicine practice, that is: alcohol, cannabis, ketamine, LSD, opiates, PCP, all these substances can induce dissociative states (depersonalization, derealization, etc.). Surprisingly, most of the hypotheses advanced so far have been formulated without reference to neuroscience. However, from a biological perspective, substance abuse among schizophrenic patients appears paradoxical: while the positive symptoms of schizophrenia might involve an hyperactivity of the reward system, the drugs of abuse all seem to increase dopamine release in that same system. That very paradox further casts some doubt on the self-medication hypothesis. And it opens an alternative: schizophrenic patients might be biologically vulnerable to the rewarding effects of drugs abuse. On the therapeutic level finally, the authors argue that polypharmacy medications such as clozapine and quetiapine, known to act on the reward system preferentially to the extrapyramidal system and known to dissociate fastly from the dopamine-D2 receptor, could simplify clinical intervention.