The purpose of this study was to investigate the diagnostic validity of schizophrenia in the Finnish Hospital Discharge Register (FHDR) with a large, epidemiologically representative sample using a multidiagnostic approach (DSM-III-R, DSM-IV, ICD-10), and to find additional criteria that could be used to improve the validity of schizophrenia diagnosis in future register-based research that utilizes the FHDR. The study population consisted of all individuals (n=877) who were born in Helsinki, Finland, between 1 January 1951 and 31 December 1960, and who had had at least one diagnosis of schizophrenia, schizophreniform disorder or schizoaffective disorder in the FHDR. All their available hospital case notes were collected. The total number of subjects for whom case notes were obtained was 806. We used the OPCRIT system (version 3.4) to produce diagnoses according to ICD-10, DSM-III-R and DSM-IV criteria based on the information extracted from the hospital case notes. We examined the distribution of the DSM-III-R, DSM-IV and ICD-10 diagnoses generated by the OPCRIT and calculated the proportion of individuals who received the same diagnosis in the FHDR and in the OPCRIT assessment. The proportion of subjects who received a core schizophrenia spectrum diagnosis (schizophrenia, schizoaffective disorder or schizophreniform disorder) in both the FHDR and OPCRIT assessment varied between 75% (DSM-III-R criteria) and 78% (ICD-10 criteria). Of the subjects with a narrow schizophrenia diagnosis in the FHDR, between 74% (DSM-IV) and 78% (ICD-10) received a diagnosis of schizophrenia in the reassessment depending on the diagnostic criteria applied. Eighty per cent of those who had received a core schizophrenia spectrum FHDR diagnosis after 1982 (vs. 56% of those who had received their last schizophrenia diagnosis in 1982 or before) received a DSM-IV diagnosis of core schizophrenia spectrum disorder. Of the 58 subjects in the sample who had been given at various times diagnoses of both core schizophrenia diagnosis and bipolar I diagnosis in FHDR, 43% received a core schizophrenia spectrum diagnosis according to DSM-IV criteria. The validity of the FHDR schizophrenia diagnosis is acceptable for large-scale register studies and comparable with that of other Nordic registers. Diagnostic validity can be further improved by selecting subjects who have core schizophrenia spectrum disorder as the latest diagnosis, by omitting cases diagnosed before 1982, and by excluding cases with a register diagnoses of both a core schizophrenia spectrum and bipolar I disorder.