The Talairach atlas of the brain was first published in 1988. The atlas is filled with highly detailed and well labeled brain sections for all three dimensions, but it also has a unique feature--a coordinate system that allows someone studying a brain to align their brain into the space described in the atlas, making it possible to locate anatomical structures using Talairach coordinates.
The coordinate system is based on three brain structures. To align a brain in Talairach space, the anterior and posterior commissures must be in a horizontal line. A line perpendicular to this, extending from the anterior commissure through the medial saggital plane, defines the vertical dimension. The axis for the third dimension is essentially the line of the anterior commissure, extending from either side of the brain to the midpoint of the commissure. The sections themselves are labeled in terms of distance (in millimeters) from the origin, which is the intersection of all axes at the midpoint of the anterior commissure.
The atlas itself is divided into four sections. The first describes how to use the atlas for functional analyses and as an aid for determining what brain structures might be affected in a damaged area. The remaining sections are divided according to what type of sections they show. The saggital sections are vertical sections (i.e. if you were standing up, the plane of the saggital sections would be perpendicular to the ground) sliced from the center of the brain out to the side. The verticofrontal sections are vertical sections made from the front to the back of the brain. The horizontal sections are displayed from top to bottom of the brain.
There are some problems with the Talairach atlas. First of all, the atlas is based on one brain, which happens to be the brain of a 60-year-old French woman. Only one hemisphere was mapped, assuming that the hemispheres were symmetrical (although as a rule they are not). The most notable difference between the Talairach brain and other brains is its size. The Talairach brain is considerably smaller than the average brain by up to 10 millimeters in each dimension. Because brain sections are typically taken every 2 millimeters, this leaves a total of 15 slices that would go unnaccounted for. Also, the Talairach atlas leaves out the majority of the brain stem and cerebellum. However, part of this problem can be solved by making an equation that uses the size ratios to transform coordinates from your brain into the Talairach coordinates. The Talairach atlas also has the advantage of being highly detailed. This is not possible in an atlas created by averaging the structures of a bunch of brains, since there is a good deal of variation from brain to brain (in fact, the surface of an "average" brain is almost completely smooth because the slight variations in locations of gyri and sulci all average out to a flat surface. It's creepy). The Talairach atlas is most accurate for areas around the origin of the coordinate system -- for example the basal ganglia, pituitary gland, and thalamus. Also, there has been extensive work comparing the Talairach atlas to other, previously developed guidelines for identifying the location of brain structures.
Happy (brain) hunting!
Talairach, J. and Tournoux, P. (1988). Co-planar stereotaxic atlas of the human brain: 3-dimensional proportional system - an approach to cerebral imaging. New York: Thieme Medical Publishers.
"Interpretation of tomographic images using automatic atlas lookup." http://www.uke.uni-hamburg.de/institute/imdm/idv/publikationen/vbc94sch/
"Anterior commissure." http://www.mrc-cbu.cam.ac.uk/Imaging/antcomm.html
"Atlases and Anatomies." for the third international conference on functional mapping of the human brain. http://orkide.lab3d.odont.ku.dk/~pet/HBM97/atlases.html
comparison of Montreal neurological institute brain to Talairach atlas. http://www.mrc-cbu.cam.ac.uk/Imaging/mnispace.html