This study provides a detailed description of the anatomical defects in the Hoxa-1-/- mutant mice previously generated in our laboratory (T. Lufkin, A. Dierich, M. LeMeur, M. Mark and P. Chambon, 1991; Cell 66, 1105-1119). Three-dimensional reconstructions of the Hoxa-1-/- rhombencephalon reveals that it bears only five rhombomeric structures (ie. morphological segments) instead of the normal seven. The first three of these rhombomeres appear normal as judged from the distribution pattern of CRABPI transcripts in the neurectoderm and from the histological analysis of the cranial nerve components derived from these structures. In contrast, the neural-crest-cell-free region normally located opposite rhombomere 5 is lacking in Hoxa-1-/- embryos, and motor neurons of the facial and abducens nerves, which normally differentiate within rhombomeres 4, 5 and 6, are missing in Hoxa-1-/- fetuses. These morphological data, combined with the determination of the molecular positional identities of the rhombomeres 4 and 5 (P. Dollé, T. Lufkin, R. Krumlauf, M. Mark, D. Duboule and P. Chambon, 1993; Proc. Natl. Acad. Sci. USA, in press), suggest that rhombomere 4 is markedly reduced, whereas rhombomere 5 is almost absent. Thus, the remnants of rhombomeres 4 and 5 appear to be fused caudally with rhombomere 6 to form a single fourth rhombomeric structure. Moreover, the migration of neural crest cells contributing to the glossopharyngeal and vagus nerves occurs in a more rostral position, resulting in abnormalities of these cranial nerves, which were visualized by whole-mount anti-neurofilament immunostaining. The mutual relationship along the rostrocaudal axis between the otic pit and the neuroepithelial site of int-2 protein secretion (a putative otogenic cue) is not significantly changed in Hoxa-1-/- embryos. However, the abnormal relationship between the rhombencephalon and the epithelial inner ear may account for the aplasia and faulty differentiation of the membranous labyrinth, the disruption of the cartilaginous otic capsule and the disorganisation of some middle ear structures. This phenotype is compared with that of the Hoxa-1-/- mutants generated by O. Chisaka, T. S. Musci and M. R. Capecchi, 1992 (Nature 335, 516-520) and with that of the mice homozygous for the kreisler mutation.