Phiomia (after the Ancient Greek phiom "lake", an ancient name for the Fayum) is an extinct genus of basal elephantiform proboscidean that lived in what is now Northern Africa during the Late Eocene to Early Oligocene some 37–30 million years ago. The type specimen of Phiomia, part of the mandible (lower jaw), was described in 1902 by Charles William Andrews and Hugh John Llewellyn. Unsure of its identity, they assigned it, tentatively, to the obsolete order Creodonta. Subsequently, it was recognised as a proboscidean. Briefly it was treated as a junior synonym of Palaeomastodon, but the two are regarded as separate genera. Though five species have been assigned to Phiomia over the years, only two, P. serridens (the type species) and P. major, are currently recognised.

Phiomia was fairly small in terms of body size, with an estimated shoulder height 134.5 cm (4.41 ft) in the case of P. serridens. In some regards it resembled Palaeomastodon, though was less basal and bore similarities to gomphotheres, to the point where it was briefly considered their ancestor. A retracted naris (nasal cavity) with strong muscle attachment sites, long snout and protruding mandible all suggest that Phiomia was among the first proboscideans to possess a true trunk. Both the upper jaw and mandibles were tusked, with those of the upper jaw being thin, recurved and blade-like, while those of the mandibles were flat, straight and broad.

The type specimen of Phiomia, a partial left mandible (lower jaw), was recovered from strata belonging to the Jebel Qatrani Formation, part of the Eocene-age Fayum fossil deposits of Egypt. In 1902, the mandible was described by Charles William Andrews and Hugh John Llewellyn Beadnell, as part of a paper naming several mammal genera from the Fayum. They assigned to it the binomial name of Phiomia serridens, after the locality of its discovery (the name "Fayum" derives from the Greek phiom, meaning "lake") and its serrated lower incisors. Andrews and Beadnell were uncertain of Phiomia's relationships, though, believing it to have been carnivorous, tentatively suggested that it was a highly specialised member of the now-disused order Creodonta. Following a review by Max Schlosser, in 1906, Andrews re-examined P. serridens, and suggested a closer relationship to Palaeomastodon, an early proboscidean. He tentatively suggested that it may have represented a smaller species of that genus, or that it belonged, at the very least, to the same family, Palaeomastodontidae. In 1908, Andrews wrote a paper on Palaeomastodon. By that point, he had become unconvinced that Phiomia could be reliably separated from it, and suggested due to its small body size that its holotype belonged to the same taxon as Palaeomastodon minor. Thus, he formally synonymised the two.

The proposed synonymy between Phiomia and Palaeomastodon lasted only fourteen years. In 1922, Matsumoto Hikoshichirō once again separated the two, based on various characteristics of the skull and teeth. Notably, the morphology of the two genera's cheek teeth (the premolars and molars) were different, with those of Palaeomastodon being bunolophodont (having inner cusps that are blunt cones, and outer cusps modified into ridges) and those of Phiomia being conventionally bunodont. Furthermore, based on a second mandible from the Jebel Qatrani, Matsumoto described a second Phiomia species, P. osborni, and reassigned two species of Palaeomastodon, thus adding P. minor and P. wintoni to Phiomia. An additional Phiomia species, P. major, was described in 2004 by William J. Sanders, John Kappelman, and D. Tab Rasmussen, based on teeth from the Chilga district of Ethiopia. In the paper describing it, the authors synonymised all of Matsumoto's taxa with P. serridens, thus reducing Phiomia to just two species. Subsequent authors have maintained this synonymy.

Since its reclassification as a proboscidean, Phiomia has always been regarded as a fairly basal member of the order. Henry Fairfield Osborn believed that it was an early member of a long-jawed mastodont lineage, and that it was directly ancestral to Trilophodon (now a synonym of Gomphotherium). In Pascal Tassy's 1988 paper on the phylogeny of proboscideans, wherein he erected the suborder Elephantiformes, he classified Phiomia as a basal member of the group, intermediate between Palaeomastodon and the later Hemimastodon. In 1992, the family Phiomiidae was erected to encompass Phiomia. A phylogenetic analysis of basal proboscideans was performed by Lionel Hautier et al. (2021), which recovered Phiomia in a similar position.

Below is a cladogram depicting the results of Hautier et al. (2021):

Phiomia serridens is estimated to have had a shoulder height of 134.5 cm (4.41 ft). P. major seems to have been larger, though by how much is uncertain due to its nature as a taxon known solely from teeth.

The skull of Phiomia is similar to that of Palaeomastodon. It was fairly long in comparison to its width, more so than in that genus, though not to the extent of gomphotheres like Gomphotherium and Megabelodon. The naris (nasal cavity) was retracted, to the extent where its topped just before the orbits. That it is surrounded by strong muscle attachment sites suggests the presence of a small trunk. Phiomia's palate was fairly narrow, while that of Palaeomastodon was wider. The symphysis of the mandible was very elongated, though again, not to the extent of the aforementioned gomphotheres; both genera were among the first proboscideans to meaningfully elongate the mandibular symphysis and tusks. Like Palaeomastodon, Phiomia had a high occiput with a large nuchal crest, possibly to counterbalance the weight of the tusks, elongated mandible, and trunk.