We aimed to estimate the value of structured interviews, medical records and Swedish register diagnoses for assessing lifetime diagnosis of patients with schizophrenia. Psychiatric records and diagnostic interviews of 143 Swedish patients diagnosed by their treating physician with schizophrenia and related disorders were scrutinized. Based on record analysis only, or a combined record and interview analysis, DSM-IV diagnoses were obtained by the OPCRIT algorithm. Independent of the OPCRIT algorithm, a standard research DSM-IV diagnosis, based on both record and interview analysis, was given by the research psychiatrist. Concordance rates for the different psychosis diagnoses were calculated. DSM-IV diagnoses based on records only, showed a good to excellent agreement with diagnoses based on records and interviews. Swedish register diagnoses displayed generally poor agreement with the research diagnoses. Nevertheless, 94% of subjects sometimes registered with a diagnosis of schizophrenic psychoses (i.e. schizophrenia, schizoaffective psychosis or schizophreniform disorder) displayed a standard research DSM-IV diagnosis of these disorders. For patients in long-term treatment for schizophrenia in Sweden, psychiatric record reviews should be optimal, cost effective and sufficient for assessment of lifetime research diagnoses of schizophrenia. For these patients a research interview adds little new information. The results further indicate that a Swedish register diagnosis of schizophrenic psychoses has a high positive predictive power to a standard research DSM-IV diagnosis of the disorders. It is concluded that for future Swedish large-scale genetic studies focusing on a broad definition of schizophrenia, it would be sufficient to rely on the Swedish register diagnoses of schizophrenic psychosis.