There is a general consensus that diagnoses for mental disorders should be based on criteria and algorithms as given in ICD or DSM. Standardized clinical interviews are recommended as diagnostic methods. In ICD and DSM, much emphasis is put on algorithms, while the underlying criteria get much less attention. The question is how valid are the criteria that are collected by structured diagnostic interviews. 209 patients from a cardiology inpatient unit were interviewed with the Mini International Neuropsychiatric Interview (MINI). 32 (15.3%) were diagnosed as suffering from a major depressive episode or dysthymia. Additionally, a thorough clinical examination was done by a psychiatric expert in 15 patients. The standardized diagnosis of present major depression was reaffirmed in one. In total, four patients were suffering from some kind of depressive disorder presently or life time. Two patients were suffering from anxiety disorders, two from adjustment disorders, and four from different types of organic brain disorders. Most important, there are 3 out of 15 who are not mentally ill. Our observations show that standardized diagnostic interviews cannot be used to make specific differential diagnoses, but rather catch unspecific syndromes. This is partly due to the fact that the wording, definition, and understanding of the underlying criteria is rather vague. This is an even greater problem if there is any somatic comorbidity. In the revision of ICD and DSM, a glossary of psychopathological terms and guidelines for the training of clinicians should be included.