A parsimony analysis of parts of the mitochondrial genes 12s and 16s (722 base pairs) from 43 species of the advanced snakes (Caenophidia) resulted in two most parsimonious topologies. Based on a strict consensus of these the following objectives were addressed: (1) Which groupings of caenophidian taxa can be recognized? (2) Is Acrochordus sister group to, or included in, Caenophidia? (3) Are boigine snakes (sensu stricto) a monophyletic grouping and where do these taxa belong in a broader caenophidian context? (4) What are the systematic affinities of the African egg-eating genus Dasypeltis? The phylogeny was then used to discuss: (5) the evolution of the posterior maxillary dentition and the composition of the retinal visual cells. The results reveal that, when using Boa constrictor as outgroup, Acmchordus is the sistergroup to the remainder of the ingroup, and a further eight clades are defined. Five genera of elapids did not appear to be monophyletic and a number of colubrids (sensu lato) such as Mehelya, Lycodonomorphus, Lampmphis, Atractaspis, and Buhoma (formerly Geodipsas), which have traditionally been problematic to place systematically, did not group with any of the larger clades. These taxa are together with the elapids representatives of very early radiations in the evolution of the Advanced snakes. The homalopsine Enhydris enhydris appears as a sistergroup to the viperine clade (Clade 1). When plotted onto the topology the posterior maxillary dentition appear to express three, maybe four, independent origins of the Opistoglyph State. The retinal evolution also appeared very complex. The suggested very primitive placing of the Boigine snakes (sensu stricto) due t o their lack of double cones in the retina of the eye was not supported here; instead the sequence data suggests this observation to be the result of a secondary loss.