We pretend with our study to compare the DSM-III diagnostic criteria with ICD-9 diagnostic criteria for psychotic episodes and to look for the diagnostic stability for both classifications. We studied 79 first psychotic episodes following ICD-9 and retrospectively we applied the DSM-III diagnostic criteria to the same patients. The mean follow up time was 46.6 +/- 6.7 months. In the ICD-9 the schizophrenic psychoses represented the main diagnostic group, not only the day they were discharged from the hospital 31.6% but also in the follow up 39.2%. On the other hand in the DSM-III, in the discharge day the schizophreniform disorder was the more frequent diagnostic 24%, while in the follow up the schizophrenic disorder came to the first position 32.9%. In the ICD-9 there was a 15.1% of diagnostic changes and in the DSM-III nearly the double 30.3%. In both classifications the unspecified and the atypical suffered a great diagnostic mobility; in both nosologies the schizophrenia was the most unchangeable diagnosis, no patient discharged with this diagnosis changed to another in the follow up. ICD-9 could be said that has a great sensitivity and an acceptable specificity for this entities, and DSM-III would have a moderate sensitivity but a very diagnostic specificity for these disorders.