While accurate for when it was made, it was over 14 years ago. New evidence of creatures that we can have no real idea on behavior, color and other details are always emerging. So, there are inaccuracies.
Walking With Dinosaurs:
What was thought to be evidence for "cannibalistic Coelophysis" has been discredited.
If the related (and likely synonymous) Megapnosaurus is any indication, Coelophysis was probably nocturnal.
The early long-necked dinosaur Plateosaurus could not walk on four legs.
The pillar-limbed croc-relative Postosuchus was most likely a biped, or at least semi-bipedal, rather than an obligate quadruped.
Placerias and the Cynodont aren't reptiles in modern phylogenetic sense, but instead mammal ancestors.
There were no Cynodonts of the size depicted in the program in the late Triassicnote other than traversodonts. Given that some Mesozoic mammals reached similar sizes (such as the infamous Repenomamus), cynodonts that big in that time period aren't strictly unlikely, but unknown from the area the episode took place. This is an example of Science Marches On rather than Somewhere, a Palaeontologist Is Crying because at the time the series was produced it was assumed that cynodonts of that size did live in Late Triassic in North America. This assumption was based on the discovery of two teeth from Chinle Formationnote though these teeth were assumed to belong to traversodont cynodonts, much different from Thrinaxodon that WWD-cynodonts were based on. However, post-WWD study indicate that these teeth can't be confidently referred to Cynodontia (or any other known group of Triassic amniotes, for that matter).
Sorry, Ornitholestes, you didn't actually have that horn-thing on your nose.
Liopleurodon was only about 9-12 meters (30-40 feet) in length, rather than the 25 meter (80 foot) long juggernaut in the series (to be fair, they said it was a huge specimen, but still, they probably couldn't grow that big even then).
Biomechanical studies have shown that skim feeding (as Rhamphorhynchus is shown doing) was not possible in known pterosaurs.
Not to mention the "iguana-spike-backed Diplodocus: some researchers now argue these spike were spread across on Diplodocus' back rather than put in a single line as shown in the program.
Another amazing example: footprints from a baby bipedal sauropod have been recently found: perhaps Littlefoot and the WWD sauropodlets walked on two legs and become quadrupedal only when they grew larger! (an ancient heritage from their ancestors, the "prosauropods" such as the aforementioned Plateosaurus). However, most paleontologists are skeptical of this interpretation. Even the trackways of adult sauropods often leave just the prints from just one pair of feet, thus is even more likely about the younger ones.
This might be nothing compared to what is seeming to come: most small-sized dinosaurs had probably some sort of covering. This is a very recent theory led by the discover of the primitive herbivore Tianyulong in China: the theory is that some kind of covering was present in the last common ancestor of all dinosaurs and pterosaurs, and then it was partially lost by its largest descendents because of the Surface area to volume ratio. Some think the "spikes" on Diplodocus have the same common origin of feathers, as well as the quill of the small herbivore Psittacosaurus and even the horny bumps lined on the back of several hadrosaur mummies. See Dinosaurs for more infos about that. Whatever the case, the old "gigantic lizards" seem to have their days numbered now.
With the discovery of Yutyrannus, it seems that multiple dinosaur groups may have had feathers.
The Tapejara species featured in the series has now been reassigned to Tupandactylus.
The giant Ornithocheirus was based on a specimen now assigned to Tropeognathus. Though it was indeed very large for an ornithocheirid, the show chose high-end, improbable estimates for its stated size. In reality, the specimen probably had around an 8m wingspan.
An example of taxonomy marching on: the "American Iguanodon" from the fourth episode would probably be placed in the genus Dakotadon today.
Honestly, it's doubtful the European Iguanodon was actually Iguanodon and not, for example, Mantellisaurus or Barilium.
Enantiornithines are now known to have lacked tail fans, contrary to the Iberomesornis shown in the series.
Leaellynasaura should have a much, much longer tail. Also, some argue it needs a plumage.
Also the giant pterosaurQuetzalcoatlus is shown as a fish eater hunting prey on the wing, while we now know it was actually stork like in habits. In fact, it probably wouldn't have hesitated to eat juvenile tyrannosaurs, like the ones in the program!
It looks like another example may be approaching. It's recently been theorized that Triceratops and Torosaurus (which were featured in Death of a Dynasty as seperate genera) are actually the same animal in different growth stages. However, research on this is still ongoing and has been doubted by some recent studies.
The accompanying book briefly mentions the possibility that Anatotitan is synonymous with Edmontosaurus. As of September 2011, this is the majority view.
They did try to partially remedy all the issues by showing Walking with Dinosaurs again in 2008 with updated narration. Unfortunately, the visuals remained untouched, so the small carnivore Ornitholestes still had a horn, coelurosaurs were still scaly, so on and so forth.
Walking With Beasts:
This series has Andrewsarchus, known only from the skull and a few fragments of bone. At the time the series was produced it was assumed to be closely related to mesonychids, and thus in the series it was modeled after mesonychids. However, laterphylogenetic studies indicate that it might have actually been a close relative of entelodonts.
However, the most likely alternative is that it was a primitive, arboreal pangolin with no armor. This makes sense (since anteaters originate in South America while pangolins appear in Eurasia, and other pangolins are known in Europe at this time) but also means that Eurotamandua, in the flesh, would look very much like a tamandua even if it wasn't a real tamandua. The use of a tamandua as a stand-in should be perfectly excusable. The use of a coati as a stand-in for the giant platypus Steropodon in WWD, on the other hand... not so much.
It now seems Gastornis was actually an herbivore, meaning its entire portrayal in the episode is completely wrong.
Walking With Monsters:
The Giant Spider in the Carboniferous was based on Megarachne, which ultimately turned out to be eurypterid ("sea scorpion") rather than spider.
The lineage that gave rise to mammals split to the one that gave rise to reptiles and birds before those invented the reptilian scales. The show represents perhaps the first time that Dimetrodon and its herbivorous "twin" Edaphosaurus have skins similar that of modern hairless mammals, instead of the classic scaly one. However, some think now that they would have the skin texture of a salamander, and the belly of a fish.
Interesting to note that the giant, Angry Guard Dog-looking Gorgonopsid from the show has scent glands (a typical mammalian feature).
The armoured plant-eating near-reptile Scutosaurus wasn't probably the ancestor of turtles. Recent research suggests that the latter were closer to modern reptiles than to Scutosaurus.
Chased by Dinosaurs:
The special Land of Giants portrayed the largest land animal of all time, Argentinosaurus, being hunted by the largest land predator, Giganotosaurus. At least one, and possibly both have since been supplanted; not long after, new evidence found that, in fact, Spinosaurus was the biggest land predator, and, although the findings are sketchy at best, Bruhathkayosaurus may be the largest land animal of all time. At the time of the show's airing, however, they were thought to be record holders.
The enormously long-necked Tanystropheus was potrayed as capable of losing and regenerating its tail like a lizard. In the past it was indeed suggested by palaeontologist Rupert Wildnote who also thought that Tanystropheus was closely related to lizards - nowadays it's generally considered to be more closely related to archosaurs than to lizards that this creature was capable of autotomy, but other scientists who studied its fossils didn't find evidence for that. It has also been portrayed as an accomplished swimmer, but we don't know for sure if it really was such - its body-shape was all but hydrodynamic, and some think Tanystropheus was a shore animal who used its neck as a fishing rod, catching small prey a bit like a heron.
Page 122 claims that therizinosaurs are known from "a lone species" from North America, probably referring to Nothronychus. Enter the ancestral therizinosaur Falcarius in 2005...
To quote page 125, "Scientist cannot agree on whether Mononykus was a bird or a [non-bird] dinosaur." The 2010 discovery of the ancestral alvarezsaur Haplocheirus confirms that Mononykus and other alvarezsaurs were not birds.
Walking with Dinosaurs 3D:
Just after they finished the Gorgosaurus models, Yutyrannus (a feathered tyrannosaur) was found.
However it should be said that this does not mean that the Gorgosaurus was inaccurate; Yutyrannus was a more primitive tyrannosaur, and only distantly related to Gorgosaurus.
It should also be noted that phylogenetic bracketing says that Gorgosaurus and kin were as likely feathered as, say, Utahraptor. And adding Tianyulong and the as-of-yet unnamed feathery Russian ornithschian, it's becoming more plausible that feathers were the original dinosaur skin, and not full-out lizard scales or crocodile scutes.
Mere days before the movie premiered, it was discovered that Edmontosaurus had a small fleshy crest on its head.