2003 was just too early for this taxon to be properly nested.
Sinodelphys (Luo et al., 2003) was considered the oldest known metathere (= marsupial) and was compared with Didelphis, the extant Virginia opossum. Here in the large reptile tree (LRT, 1250 taxa, subset Fig. 1) Sinodelphys nests between Chaoyangodens and Brasilitherium + Kuehneotherium among the prototheres, basal egg-laying mammals. Sinodelphys may have been mistakenly nested because Chaoyangodens and Brasilitherium are newer taxa. Several of the other taxa are also more recently published.
Figure 2. Mesozoic prototherians + Megazostrodon, the last common ancestor of all mammals. Only two taxa (gray) are post-Cretaceous.
The Luo et al. study nests Sinodelphys
just inside the Metatheria, very close to the Eutheria/Metatheria split. Among taxa both analyses have in common, very few have matching sister taxa. Many are not even in the same large clade (Eutheria/Metatheria/Prototheria). This may be due to an over reliance on dental traits in the Luo et all. study and an under reliance of dental traits in the LRT, which employs a wider gamut of taxa (vs. taxon exclusion in the Luo et al. study).
Figure 2. Sinodelphys skeleton in situ with select bones colored using DGS.
Clearly
Sinodelphys has a dorsal naris with short ascending processes on the premaxilla, not a terminal naris opening anteriorly. This trait alone nests Sinodelphys with the egg-laying mammals. Even so, a long list of traits support that nesting. Perhaps if Sinodelphys were described today, after so many other prototheres have been reported, it would have been identified as one of them.
Figure 3. Skull and forelimbs of Sinodelphys in situ. Arrow shows the displacement of the entire hand that otherwise appears to be lost beyond the matrix. How fortuitous!
With an inch-long skull
this is a tiny Early Cretaceous egg-layer, ancestral to today’s platypus and echidna.
Figure 4. Reconstruced skull of Sinodelphys based on DGS methods. This is very close to Brasilitherium, but with a larger set of canines. Like other prototheres, the nares are dorsal, not terminal.
The fingers on both hands are jumbled up (Fig. 3).
If Luo et al. are correct in their manus reconstruction, the only change I would make is to flip it left to right. Note their digit 5 is missing the proximal phalanx (Fig. 5). That is more likely the thumb because then digits 3 and 4 are the longest, as in sister taxa in the LRT.
Figure 5. Manus of Sinodelphys as originally reconstructed. Flipping the hand, as in the revised image, more closely matches sister taxa with digits 3 and 4 the longest.
References
Luo Z-X, Ji Q, Wible JR and Yuan C-X 2003. An Early Cretaceous tribosphenic mammal and metatherian evolution. Science 302:1934–1939.
The Pterosaur Heresies 