Clozapine does not constitute a first-line treatment due to the occurrence of agranulocytosis. However, the benefit risk/ratio fully justifies its use in two situations: resistance to neuroleptics, intolerance to neuroleptics. There are no internationally recognized objective criteria to define resistance. The ones defined in the well-restricted methodological environment of a clinical trial are generally not applicable to daily practice. In particular, the accepted criteria do not always take into account the personal factors, the social and environmental context and rehabilitation programs. May et al. for example defined in 1988 the degree of response to therapy based on clinical improvement as well as social integration. It is widely recognized that if the severity of residual symptoms (grade 5 response) requires the hospitalization, treatment with clozapine is warranted. For partial responders to classical treatment (grade 4 response) who could benefit from clozapine, the risk ratio of clozapine needs to be further evaluated. The identification of predictive factors of response to therapy would allow a novel approach of this indication. According to certain authors, paranoid type responds best to therapy. However, the evidence collected to date needs to be confirmed. In particular, the effects of clozapine on predominantly negative symptoms require further investigations. In contrast to several european studies, intolerance to neuroleptics is rarely a reason for initiation of clozapine therapy. This would indicate that the appreciation of intolerance to neuroleptics notably varies from one country to the next. Intolerance criteria need to be further specified as well as the benefit risk/ratio.(ABSTRACT TRUNCATED AT 250 WORDS)