The contrast between the great importance of psychiatric disorders and the fact that they are often not identified as thus in medical and surgical units has been emphasized during the last ten years. The prevalence of depression disorders in an hospitalised population of medical patients is generally estimated between 15% and 30%. It is admitted that one third to two third of theses cases are not diagnosed by non psychiatrist physicians. It is the reason why self questionnaires have been finalized in order to detect psychiatric troubles in medical populations. The General Health Questionnaire (GHQ), constructed by Goldberg, is surely the most internationality used, but no validation study has yet been undertaken in France. This study means the validation of a french version of GHQ-28 items, set on 46 diabetics (type 1 and type 2) selected in a week hospital. This validation is based on the comparison of the two principal scoring methods: "GHQ" (0,0,1,1); "CGHQ" (0,1,1,1 for negative items) and the determination of the best cut-off score in this specific population. The referring DSM III-R diagnoses were assessed by the Composite International Diagnostic Interview (CIDI; Robins et al., 1988), which is a highly structured psychiatric interview developed by the World Health Organisation. Sixteen patients out of the 46 diabetics showed criterions of depressive and/or anxious disorders, 6 of them presenting both. The analysis of the answers to the questionnaires was based upon the technic of the Relative Operating Characteristic (ROC) curves which takes all the sensitivity-specificity couples bound to each of the possible threshold-score into account.(ABSTRACT TRUNCATED AT 250 WORDS)