A re-appraisal of <i>Parvicursor remotus</i> from the Late Cretaceous of Mongolia: implications for the phylogeny and taxonomy of alvarezsaurid theropod dinosaurs

<i>Parvicursor remotus</i> Karhu &amp; Rautian, 1996, based on a fragmentary skeleton from the Campanian Barungoyot Formation at Khulsan, in the Gobi Desert of Mongolia, is re-described in detail. Contrary to the original description, we find no evidence that the last dorsal vertebra is opisthocoelous. A biconvex last dorsal vertebra could be present in <i>Parvicursor</i> as in other parvicursorines. The transversely compressed sacral centrum with a sharp ventral keel is the last sacral, not the first sacral as originally alleged. The pelvic bones are not fused at the acetabulum. The holotype of <i>P. remotus</i>, the smallest known alvarezsaurid specimen, is a juvenile individual far from skeletal maturity. New morphological details of <i>Parvicursor</i> that are revealed include a ventral ridge on the posterior dorsal vertebrae; a horizontal supraacetabular crest of the ilium whose anterior end is placed anterior to the most ventrally projecting part of the pubic peduncle; the possible presence of a pubic obturator foramen; an entocondylar tuber on the femur; internal flexor flanges on metatarsals II and IV; a lateral flexor ridge on metatarsal IV; and a proximoventral notch on pedal phalanx IV-1. The Tugriken Shireh alvarezsaurid is not referable to <i>Parvicursor</i> because of its much larger ungual phalanx on the second pedal digit. <i>Kol ghuva</i> is excluded from Alzarezsauridae. The two successive sister taxa for <i>Parvicursor</i> are <i>Ceratonykus</i> and <i>Linhenykus</i>. <i>Ceratonykus</i> differs little from <i>Parvicursor</i> and might be a synonym of the latter. <i>Linhenykus</i> is very similar to <i>Parvicursor</i> in all overlapping skeletal elements. The Parvicursorinae includes all Asian Late Cretaceous alvarezsaurids except <i>Qiupanykus</i> and <i>Nemegtonykus</i>.