The earliest evidence for animal life comes from the fossil record of 24‐isopropylcholestane, a sterane found in Cryogenian deposits, and whose precursors are found in modern demosponges, but not choanoflagellates, calcareans, hexactinellids, or eumetazoans. However, many modern demosponges are also characterized by the presence of siliceous spicules, and there are no convincing demosponge spicules in strata older than the Cambrian. This temporal disparity highlights a problem with our understanding of the Precambrian fossil record – either these supposed demosponge‐specific biomarkers were derived from the sterols of some other organism and are simply retained in modern demosponges, or spicules do not primitively characterize crown‐group demosponges. Resolving this issue requires resolving the phylogenetic placement of another group of sponges, the hexactinellids, which not only make a spicule thought to be homologous to the spicules of demosponges, but also make their first appearance near the Precambrian/Cambrian boundary. Using two independent analytical approaches and data sets – traditional molecular phylogenetic analyses and the presence or absence of specific microRNA genes – we show that demosponges are monophyletic, and that hexactinellids are their sister group (together forming the Silicea). Thus, spicules must have evolved before the last common ancestor of all living siliceans, suggesting the presence of a significant gap in the silicean spicule fossil record. Molecular divergence estimates date the origin of this last common ancestor well within the Cryogenian, consistent with the biomarker record, and strongly suggests that siliceous spicules were present during the Precambrian but were not preserved.