The unbiased sampling techniques of stereology have been developed to avoid the inaccuracies of using 'representative' sections for morphometric studies. In order to carry out a stereologically valid study, the region of interest must be fully available for sampling and its boundaries or its constituents must also be distinctly identifiable. However, in the setting of a brain bank in which only one half of the brain specimen is fixed for morphology, a logistic problem arises in satisfying the needs of the diagnostic neuropathologist with that of the stereologically oriented morphologist. We present a dissection approach in which the region for analysis must be used for unbiased sampling and also be available for paraffin-embedded neuropathologic work-up. Following fixation, a block consisting of the entire region of interest is removed intact and using a multibladed knife the block is subsectioned in the coronal plane at regular intervals. Alternate blocks are chosen for either paraffin embedding and destined to neuropathologic evaluation or are processed for stereology. The stereology blocks can be either cryoprotected or placed in phosphate buffer and are serially sectioned on a cryostat or a Vibratome. Preliminary analyses using this approach have provided reliable estimates of the total number of different neuronal populations and disease-related lesions in a variety of human and non-human banked specimens. In addition, this approach has definite advantages in that it provides rigorous quantitative estimates of neuropathologic changes that can be correlated to clinical data and does not compromise the routine neuropathological diagnostic procedure of the materials.