NeuroNames is a nomenclature designed as a tool for indexing digital databases of neuroscientific information. It can be used, for example, as the entry point to a digital dictionary of neuroanatomy, to a brain atlas, or to a database of information referenced to specific brain structures. The user can query with terms from many different nomenclatures. One can create a neuroanatomic ontology from NeuroNames by relating an appropriate subset of terms to a conceptual model represented by structures illustrated in a brain atlas. At the conceptual core of NeuroNames are primary structures, the elementary units of the brain in the spatial domain. Each primary structure is associated with a set of synonyms that represent the structure in the symbolic domain. One of the synonyms is designated the default name for use in verbal definitions of other structures. A unique abbreviation based on the default name is provided for labeling images. Neuroscientists classify structures in different contexts reflecting different attributes of interest. Thus, the name of a given structure can appear in any number of hierarchical contexts. In NeuroNames all primary structures are now represented in at least two hierarchies. The first is a nine-level "Brain Hierarchy," in which volumetric structures are grouped by proximity to form successively larger units that represent the brain at different levels of dissection. Secondly, primary structures are categorized in a three-level "spatial attribute hierarchy" used to color- code them for visual display. Grouped structures in the nine-level volumetric hierarchy are designated superstructures, each of which has synonyms, a default term, and an abbreviation. All names of structures not in the hierarchy are designated ancillary terms and are defined in words using the default names of hierarchy structures. With NeuroNames as entry point, we have developed BrainInfo (http://braininfo.rprc.washington.edu), a website that allows searchers to proceed intuitively in a few steps to descriptions and images of specific structures. Currently NeuroNames resides in a Microsoft ACCESS database and includes some 12,200 terms in seven languages.