Serial section reconstruction shows that kinetid ultrastructure in two genetically divergent Paracercomonas (P. virgaria, P. metabolica) is basically similar, differing somewhat from clade A cercomonads. Paracercomonas (Paracercomonadidae fam. n.) have a posterior root (dp1) attached to the posterior centriole, unlike Cercomonadidae (here revised to include only Eocercomonas, Cercomonas, Filomonas gen. n., and Neocercomonas), which belong in clade A (new suborder Cercomonadina) with Cavernomonas (Cavernomonadidae fam. n.). Whether dp1 is serially homologous to anterior root da is unclear. The common ancestor of Cercomonadida probably had five microtubular roots, two fibrillar microtubule-nucleating centres generating microtubular cones, and striated connectors between obtusely angled centrioles. Our new data leave the question of holophyly versus polyphyly of Cercomonadida unresolved, but clarify cercozoan root diversity and homologies. Ventral root vp1 is throughout Cercozoa; vp2 might be restricted to the new superclass Ventrifilosa plus Sarcomonadea. Though cercozoan microtubular arrangements differ substantially from others within the kingdom Chromista, the microtubular root numbering system used for other chromists and Plantae is applicable to them; in doing this we found that the single anterior root of excavates (probably ancestral to Chromista, Plantae and unikonts) and Euglenozoa corresponds with R3 (not R4 as previously thought) of corticate eukaryotes (Chromista plus Plantae).