All posts

Methodology

How to Check an Animal Hospital Patient Before You Admit It

Published 2026/07/14

Last updated 2026-07-14.

First, the disclosure that shapes this whole post: we do not play Animal Hospital. Everything below is read from the public community write-ups — Beebom, Tech Nerdiness, fdaytalk, BloxGuidesGG — read against each other. It is a reconstruction, not an in-game audit. Where the write-ups disagree, this post says so instead of picking a winner.

With that out of the way: the interesting thing about the anomaly list is not the tells. It is the order you check them in, because the order decides whether a tell can reach you at all.

The order is the whole trick

Look, photograph, monitor, decide.

The sequence matters because the three categories hide in three places that do not overlap. An appearance tell is visible at the glass. A photo tell is invisible at the glass by design — that is the entire point of it. A CCTV tell is invisible at the glass and in the photo.

So a patient that survives your eyes has told you almost nothing about what the camera will show. Seven of the eighteen tells we track only appear on the feed. "It looked normal" is not evidence; it is the absence of one of three kinds of evidence. The only thing that justifies opening the door is: the window was clean, the photo matched, and the feed was clean.

Check one: the window

Six tells live here — Three Eyes, Hollow Face, Human Teeth, Creepy Smile, Googly Eyes and Sharp Teeth, and Twitching.

Most are loud. Three Eyes is described as three slightly mismatched glowing red eyes rather than two, sometimes with the original pair visible underneath, and it is the most-cited "obviously an anomaly" sign across every write-up. Hollow Face is empty sockets and a blank, open-mouthed expression.

The quiet one is Human Teeth: the patient behaves completely normally, and the only thing wrong with it is realistic human teeth where animal ones should be. That is the argument for checking the mouth on every patient rather than only the ones that already look odd — by the time a patient looks odd, you did not need the reminder.

One correction worth making loudly, because we had it wrong ourselves until we re-read the sources: Head Banger is not an appearance tell. Every write-up we checked files it as a hazard or a monster. It bangs its head on the glass, interacting with it unprepared drains sanity, and the fix is coffee or waiting it out — but you never decide whether to admit it, so it does not belong in a screening routine at all. The checker no longer offers it as a sign.

Check two: the photo

Five tells live in the developed photo, and this is the step players skip most, precisely because the patient in front of them already looked fine.

  • Different In Photo / Incorrect Photo — the print shows a different model, different markings, or an entirely different animal.
  • Different Eyes — the eyes in the print do not match the patient: changed shape, an added eyelid, a missing pupil.
  • Unnatural Photo — the patient looks like a normal cartoon animal at the window, but the print shows an eerie, unnaturally realistic face.
  • Cursed Photo — bloodshot eyes, a grotesque smile, blurred features. Picking it up before it finishes developing costs 10 sanity, so let it finish, then look.
  • Photo Static — and here the write-ups actively disagree.

On static: BloxGuidesGG says static with no other tell does not guarantee a Shapeshifter but is still an anomaly. fdaytalk says that as of June 2026 static alone is no longer conclusive, and describes testing the patient rather than rejecting on static by itself. We cannot tell you which is current, so treat static as a reason to check everything else very carefully rather than as a settled reject.

Where the write-ups do converge is on method: make the photo comparison mechanical. Eyes, then ears, then colour, then shape — four discrete looks. Asking yourself whether the print "feels" different reliably produces a no under pressure, because the differences described are as small as one missing pupil.

Check three: the cameras

Seven tells only exist on the feed — Void Body, Deformed Body, Skinwalker, Different Ears, The Starer, Black Eyes, and CCTV Twitching. That is nearly half the catalogue sitting behind a step that takes a few seconds.

Void Body is the freebie: the whole body goes solid black on camera. Nothing to interpret.

Skinwalker — other guides call it the Disguised Shapeshifter — is the one that punishes you. It looks completely normal at the window and in the photo, then shows a wide-open mouth full of sharp teeth on the feed once it reaches check-in. The write-ups agree on a specific, actionable detail: use the lobby camera for a side-on view, because the check-in camera can hide the tell entirely.

Deformed Body stretches and distorts, and the distortion changes shape each time you switch angle or reopen the feed. That change is the confirmation — a normal patient does not look different between two looks.

Different Ears needs a correction we owe you. We used to describe it as camera-only. It is not: the write-ups place mismatched ears in the photo as well as on the feed. Check ears twice, once against the print and once on the camera. If you only ever hunt this one on the cameras, the photo version walks straight past you.

The Starer has a confirmation step built in — it always faces the camera, and if you switch feeds it turns to look at that one too.

Two rules that survive contact with panic

One matched sign is enough. Nothing in the list turns out to be a normal patient once a tell is confirmed, so hunting for a second one only spends time you do not have.

Do not promote a vibe to a verdict. Under pressure it is tempting to invent a rule mid-shift — "that one moved oddly." If nothing in the list matches, finish the remaining checks instead of acting on a feeling.

That is not a claim that our list is complete. It is not, and the write-ups name tells we could not pin down well enough to publish as fact — we list those separately rather than guessing. If you see something repeatable that is not in the list, trust your eyes and tell us.

That gap is what the anomaly checker exists to close: click the sign you noticed and read the call, so you are not relying on memory at the worst possible moment. It is a memory aid, not an oracle. A clean result means nothing on our list matched what you told it — not that the patient is safe.

FAQ

Can I skip the CCTV check if the patient obviously looks fine? No. "Obviously looks fine" is exactly the condition the camera-only tells were built for — seven of the eighteen appear nowhere else.

Is one sign really enough to reject? Yes, according to every write-up we read. Looking for a second one costs time and changes nothing.

Which tell gets missed most? The ones on a patient that reads as completely normal: Human Teeth at the window, Different Eyes and Unnatural Photo in the print, Skinwalker on the feed.

Is a static photo an automatic reject? The sources disagree, and that is the honest answer — see the photo section above.

Are there Animal Hospital codes? No. The game has no code redemption box at all. See the codes page for the current status and how we check it.