The staging method proposed by Braak and Braak is based on the sequential accumulation of neurofibrillary pathology in the cerebral cortex. Unlike the currently used diagnostic criteria for Alzheimer's disease (AD) it does not take into consideration the age of the patients and whether they were demented or not for the establishment of the pathological stage of the disease. To examine the interobserver reliability of the method we performed an inter- and intrarater study using the Braak staging method in 41 brains. The agreement between the examiners and between the diagnoses of the same examiner at different times was almost perfect, the kappa statistics reaching values above 0.90. These findings indicate that the staging, relying on the differential distribution of neuritic pathology in the brain in AD, is a reliable and reproducible method for the description of AD-related pathology. This makes it suitable for brain-banking and research purposes.