In 1891, two years after the naming of Triceratops, a pair of ceratopsian skulls with elongated frills bearing holes were found in southeastern Wyoming, Niobrara County, by John Bell Hatcher.
Hatcher's employer, palaeontologist Professor Othniel Charles Marsh, coined the genus Torosaurus for them. While an estimated 2,000 Triceratops specimens have been collected from the American West, only seven partial skulls of Torosaurus have been found, so they are pretty rare.
Like Triceratops, they had massive skulls. Torosaurus had one of the largest skulls of any known land animal, with the frilled skull reaching 2.77 metres (9.1 ft) in length. Torosaurus were about the same size as Triceratops.
They had an elongated frill with large openings (fenestrae), long squamosal bones of the frill with a trough on their upper surface, and the presence of five or more pairs of hornlets (epoccipitals) on the back of the frill. Torosaurus also lacked the long nose horn seen in Triceratops prorsus, and instead resembled the earlier and more basal Triceratops horridus in having a short nose horn. Three species have been named, Torosaurus latus, T. gladius and T. utahensis. T. gladius is no longer considered a valid species.
In 1933, Richard Swann Lull increased this to 2.4 metres and 2.57 metres respectively. Based on this, Torosaurus was thought, albeit briefly, to have the longest skull of any known land animal.
Sixty-five years later in 1998, Thomas Lehman claimed that a Pentaceratops specimen possessed a partial skull that would have been 2.9 metres long in life. This was again doubted by Nicholas Longrich who in 2011 named this exemplar as a separate genus Titanoceratops and concluded its skull had been reconstructed as too long.
In 2006, Andrew Farke, a palaeontologist at the Alf Museum of Paleontology in South Dakota, pointed out that the new skulls described by him were on average even longer than Hatcher's original two: MOR 1122 has a length of 252 centimetres and MOR 981 of 277 centimetres.
Farke’s research interests focus on exploring the Cretaceous continental ecosystems of North America, particularly the ceratopsian (horned) dinosaurs, with active fieldwork in California and Wyoming.
In 2006, Farke published some diagnostic traits of Torosaurus. The frill is extremely long in comparison to the remainder of the skull. The rear, parietal, and edge of the frill bear ten or more epiparietals, triangular osteoderms. A midline epiparietal is absent; likewise, no osteoderm straddles the parietal-squamosal boundary. The parietal bone is thin. It is pierced by parietal fenestrae in the form of circular or transversely oval openings. The parietal bone is about 20% wider than long.
Farke also identified a single trait in which T. latus differed from both Triceratops horridus and T. utahensis: its squamosal bore a conspicuous ridge on the edge with the parietal combined with a deep longitudinal trough parallel to it.
Farke pointed out that the known Torosaurus specimens are rather variable. The orbital "brow" horns are sometimes large and curved to the front, as with MOR 981, and sometimes short and straight as shown by MOR 1122 and ANSP 15191.
Also, the position of these horns differs: often they are located directly on top of the eye socket but with YPM 1831 they originate at the rear edge of the orbit. Likewise, there is a variation in the form of the nose horn. YPM 1831 and to a lesser extent YPM 1830 have a straight upright nasal horn but MOR 981, ANSP 15192 and especially MOR 1122 at most possess a low bump. The frill too differs. ANSP 15192 and YPM 1830 have a shield curving upwards at the rear, but the frill of YPM 1831 is nearly flat, though this could be an artefact of restoration.
The frill of YPM 1831 is also heart-shaped, with a clear midline notch, whereas the rear edge of the other specimens is straight. The frill proportions are quite variable: with YPM 1831 the length-width ratio is 1.26 but MOR 981 has a shield 2.28 times longer than wide. The number of epiparietals is difficult to assess as most fossils seem to have lost them. MOR 981 and MOR 1122 have ten and twelve epiparietals respectively. YPM 1831 has been restored with a fontanelle in the skull roof, which possibly is authentic. Farke also concluded that the degree of variability did not exceed that shown by related genera.
Farke stressed that, apart from the frill, no systematic differences could be found between Torosaurus and Triceratops. All Torosaurus specimens are similar in that they lack a truly long nasal horn and a horizontal arterial groove at the front base of that horn, but Triceratops fossils with the same combination of traits are not uncommon.
In 2008, Hunt concluded that T. utahensis, contrary to T. latus but similar to Triceratops, possessed a midline epiparietal.