Directed indels, insertions, and deletions within paralogous genes, have the potential to root the tree of life. Here we apply a newly developed rooting algorithm, top-down rooting, to indels found in informational and operational gene sets, introduce new computational tools for indel analyses, and present evidence (P < .01) that the root of the tree of life is not present in its traditional location, between the Eubacteria and the Archaebacteria. Using indels contained in the dihydroorotate dehydrogenase/uroporphyrinogen decarboxylase gene pair and in the ribosomal protein S12/beta prime subunit of the RNA polymerase gene pair, we exclude the root from within the clade consisting of the Firmicutes plus the Archaebacteria and their most recent common ancestor. These results, plus previous directed indel studies excluding the root from the eukaryotes, restrict the root to just four possible sites. One potential root is on the branch leading to the double-membrane prokaryotes, another is on the branch leading to the Actinobacteria, another is within the Actinobacteria, and the fourth is on the branch leading to the Firmicutes–Archaea clade. These results imply (1) that the cenancestral population was not hyperthermophilic, but moderate thermophily cannot be excluded for the root on the branch leading to the Firmicutes–Archaea clade, (2) that the cenancestral population was surrounded by ester lipids and a peptidoglycan layer, and (3) that parts of the mevalonate synthesis pathway were present in the population ancestral to the Bacilli and the Archaebacteria, including geranylgeranylglyceryl phosphate synthase, an enzyme thought to be partially responsible for the unique sn-1 stereochemistry of the archaeal glycerol phosphate backbone.