BACKGROUND: Since the pioneering studies of human thalamic anatomy based on histology and binding techniques, little new work has been done to bring this knowledge into clinical practice. OBJECTIVE: With the advent of magnetic resonance imaging (MRI) we hypothesized that it was possible, in vitro, to make use of high spontaneous MRI contrasts between white and grey matter to directly identify the subcompartmentalisation of the thalamus. METHODS: An anatomic specimen was imaged at high field (4.7 T) (basal ganglia plus thalamus block; 3-dimensional (3D) T1-weighted spin echo sequence; matrix, 256 x 256 x 256; isotropic voxel, 0.250 mm/edge; total acquisition time, 14 hours 30 minutes). Nuclei were manually contoured on the basis of spontaneous contrasted structures; labeling relied on 3D identification from classic knowledge; stereotactic location of centers of nuclei was computed. RESULTS: Almost all intrathalamic substructures, nuclei, and white matter laminae were identified. Using 3D analysis, a simplified classification of intrathalamic nuclei into 9 groups was proposed, based on topographic MRI anatomy, designed for clinical practice: anterior (oral), posterior, dorsal, intermediate, ventral, medial, laminar, superficial, and related (epi- and metathalamus). The overall 4.7-T anatomy matches that presented in the atlases of Schaltenbrand and Bailey (1959), Talairach et al (1957), and Morel et al (1997). CONCLUSION: It seems possible to identify the subcompartments of the thalamus by spontaneous MRI contrast, allowing a tissue architectural approach. In addition, the MRI tissue architecture matches the earlier subcompartmentalization based on cyto- and chemoarchitecture. This true 3D anatomic study of the thalamus may be useful in clinical neuroscience and neurosurgical applications.