This is a clinical study of failed suicides, in order to contribute to the explanation of completed suicides. 423 suicidal patients were admitted to the emergency critical care unit of a metropolitan municipal hospital during 6 years. The catchment area of the unit is the east four wards of Tokyo. Almost all patients who attempted suicide and four who later on completed suicide were interviewed by consultating psychiatrists. The total number of cases thus investigated was 265. Medical and psychiatric charts were retrospectively evaluated for the study. According to the lethality of respective suicidal methods, the 265 cases were divided to the ABSOLUTELY DANGEROUS GROUP (133 patients) and the RELATIVELY DANGEROUS GROUP (132 patients). The former, except for the few patients who later on died, could be considered as "failed suicides"; 75% of these patients were mentally disordered (psychoses, depressive disorders or psychoactive substance use disorders), whereas in the latter group the respective figure was 48%. The rates of each disorder were different in the age classes: younger (below 30), middle aged (30-49), and older (50 and over). In the younger group, psychoses (F2 cord of ICD-10 draft) were the main cause (52%). By contrast, endogenous major depression was the main cause in the older group (48%). The rate of psychoactive substance use disorders was highest (22%) in the middle aged among the three classes. There were 1562 officially recorded completed suicides in the catchment area during the same 6 years. From the distribution of diagnoses in each of the three age classes of the ABSOLUTELY DANGEROUS GROUP and the distribution of age classes in 1562 completed suicides, the rates of mental disorders amongst the completed suicides were estimated as roughly 26% psychoses, 46% depressive disorders and 18% substance use disorders. The total figure was 90% and quite similar to the results of previous studies by psychological autopsies in the western countries. For an attempt to decrease the rates of suicide, the most important point seems to be the prevention of suicides by the mentally disordered. Depression is most prevalent among elderly suicidal patients. In contrast to previous studies, the author found psychoses most often in younger patients. Psychotic symptoms of schizophrenic and of alcoholic suicidal patients were analysed in detail. None of them nor any other patient could be considered as a "rational suicide" case.