Background: Because type II diabetes is currently more prevalent in schizophrenics than non-schizophrenics, some investigators have postulated a more fundamental association between these two disorders. However, there is a paucity of evidence concerning the association between diabetes and schizophrenia that is not confounded by the use of antipsychotic drugs, many of which have been linked to diabetes or pre-diabetes. Clinical literature from the pre-antipsychotic era (PAPE) may provide insights that help better define the risks currently associated with antipsychotic drugs. Aims and methods: After reviewing approximately 140 articles that addressed metabolic derangements in serious mental illness, we analyzed 20 English language clinical studies from the PAPE (all pre-1950 to assure that none of the patients had been exposed to neuroleptics) that specifically commented on glucose regulation in patients with schizophrenia and severe depression. From these, data from 5 studies that used uniform methodology and presented blood glucose (BG) data en toto were selected for quantitative evaluation. Fasting blood glucose (FBG) values and results from oral glucose tolerance tests (OGTTs), corrected by a factor of 20 mg% for the presence of non-glucose reducing substances, were pooled from patients with schizophrenia, schizophrenia sub-classified as catatonic, patients with psychotic depression and age-matched controls. BG data were evaluated with reference to current diagnostic standards for diabetes mellitus (DM), impaired fasting glucose (IFG) and impaired glucose tolerance (IGT) from the American Diabetes Association. Results: The mean FBG from the schizophrenic group (n = 148; 88.1 ± 17.8 mg% (range 50-140 mg%)) was not materially different from the mean FBG of contemporaneous age-matched healthy controls (n = 27; 82.3 ± 10.4 mg% (range 58- 105 mg%); p> 0.05). Descriptions of patients by the original authors did not indicate clinical signs of DM. However, these acutely ill schizophrenics were nearly 4 times as likely as controls (relative risk 3.74) to manifest a moderate elevation in FBG not inconsistent with IFG according to current diagnostic standards (i.e., FBG 100-125 mg%), and were also approximately 3 times as likely as controls (relative risk 2.9) to have fastinghypoglycemia (i.e., FBG < 65 mg%). With the notable exception of a subgroup with active catatonia, OGTT results from schizophrenics provided equivocal evidence of IGT. Evidence is presented which suggests that, in the setting of acute or sub-acute serious mental illness, hyperglycemia was often transitory, tending to resolve spontaneously or in apparent response to non-pharmacologic psychotherapy. Summary and conclusions: Insights from studies performed in the PAPE cast doubt on the clinical relevance and putative specificity of association between diabetes or pre-diabetes and schizophrenia. Moderate fasting hyperglycemia, moderate fasting hypoglycemia and equivocally abnormal OGTTs were present in substantial minorities of schizophrenics early in the course of hospitalization, but these findings were nonspecific and appear to have been transitory in many instances. We conclude that it is inappropriate to refer to available BG data from the PAPE as supporting the hypothesis that a diagnosis of schizophrenia per se is an independent risk factor for diabetes. Additional research is needed to better define environmental and iatrogenic factors that may explain the higher rates of true diabetes and related morbidity that have been reported recently in schizophrenic cohorts, and in particular the attributable risk of antipsychotic drugs.