To simplify the understanding of schizophrenia, psychiatry has invested important efforts into the modelling of the disease. For obvious ethical reasons, sleep deprivation and sensory deprivation models have been rejected in the last decades. Thus, the psychotic disorders induced by psychoactive substances remain the main heuristic models of schizophrenia. Most of the work in the pharmacological modelling of schizophrenia has occurred in the sixties and seventies. Still, some observations retain their interest: 1) lasting up to six months, the amphetamine psychosis remains the best model of the positive symptoms of paranoid schizophrenia; 2) hallucinogens such as LSD or psilocybine reproduce, at best, the psychotic phenomenology observed during the first episodes of schizophrenia; 3) anesthetic dissociatives (PCP & ketamine) faithfully reproduce thought disorder, as well as the frontal cognitive deficits and certain negative symptoms of schizophrenia; 4) cannabis reproduces with fidelity the depersonalization states observed among schizophrenics; 5) alcohol hallucinosis could be the best model of the patients hallucinations; 6) the psychotic disorder induced by muscarinic antagonists is not usually considered as a schizophrenia model, but its occurrence signals that cholinergic dysfunctions could be associated with the cognitive symptoms of the disease. As such, these observations go hand in hand with the common view of schizophrenia as a pathology of general cerebral disconnectivity that would affect a plurality of neurotransmitter systems.