This paper describes a prospective follow-up of 364 patients initially notified as suspected Creutzfeldt-Jakob disease to a Surveillance Unit in Göttingen, Germany. Six patients were diagnosed as having genetic prion disease by blood analysis and were excluded from the study. After examination and review of the remaining 358, 193 were classified as probable Creutzfeldt-Jakob disease. However, autopsy revealed that five of the 193 did not have Creutzfeldt-Jakob disease (four cases, Alzheimer's disease; one case, cerebral lymphoma). Of the 54 patients classified as possible Creutzfeldt-Jakob disease, 10 had another diagnosis made at autopsy. Two of the 111 cases originally classified as having other diseases were found to have Creutzfeldt-Jakob disease on autopsy. Autopsy evidence, together with follow-up of the patients still living and those who died without autopsy, revealed a broad range of other diagnoses. In the younger age groups, the commonest were chronic inflammatory diseases including Hashimoto encephalitis, whilst rapidly progressive Alzheimer's disease was most common in the older age groups. The presence of 14-3-3 protein in the CSF discriminated better between Creutzfeldt-Jakob disease and other rapidly progressive dementias than did the EEG pattern or the MRI. The inclusion of this CSF protein in the criteria of Masters and colleagues (Ann Neurol 1979; 5: 177-88) improves the accuracy and confidence in the clinical diagnosis of Creutzfeldt-Jakob disease.