Dog Ate Grapes or Raisins - What the Sources Report

Your dog has eaten grapes or raisins and you want a number that tells you whether to worry. This page has one number on it, and the most useful thing about that number is that a second veterinary source of equal standing refuses to publish it.

That is not evasion. It is the actual state of the evidence in August 2026, and knowing it changes what you do next, which is ring somebody rather than sit and count fruit.

What changed, and why the old advice was so vague

For decades the honest answer to why grapes hurt dogs was that nobody knew. Owners were told the fruit was dangerous, that some dogs ate a bagful and were unharmed while others ate a handful and died, and that no mechanism explained it. That is an unsatisfying thing to be told when you are frightened.

ASPCA Pro, read on 4 August 2026 and carrying a publication date of 21 December 2022 with an update on 9 June 2025, describes how the picture changed. Clinicians connected “a case in which a dog developed acute kidney injury after ingesting homemade playdough with cream of tartar to grape poisoning in dogs”. Cream of tartar is potassium bitartrate. Follow that thread and you arrive at tartaric acid, which ASPCA Pro describes as “a nephrotoxin in dogs” and “the common denominator and the likely toxic component”. Post-mortem renal lesions in dogs exposed to cream of tartar and tamarind resembled those seen in grape and raisin cases.

The MSD Veterinary Manual, read the same day, puts the mechanism this way: dogs cannot excrete organic acids effectively, so tartaric acid accumulates in the proximal renal tubular cells. Cornell’s Riney Canine Health Center names “tartaric acid and its salt, potassium bitartrate, as the most likely cause” and adds that these compounds “are highly concentrated in grapes, raisins, sultanas”.

So the toxic principle now has a name. What it does not have is a reliable dose, and the reason is in the chemistry rather than in the research effort.

Why the amount you counted is a weaker number than it looks

ASPCA Pro gives the range that undoes the arithmetic: “The concentration of tartaric acid in grapes varies from 0.35% to 2%, which makes predicting toxicity difficult.” That is close to a sixfold spread between one grape and another. Variety, ripeness, growing conditions and whether the fruit has been dried all move it.

MSD says the same thing in its own words: “The tartaric acid content of grapes, raisins, and tamarinds varies, resulting in variable effects following ingestion.”

Now add the second uncertainty, which is yours. You are counting after the event. You know how many are missing from the bowl, not how many went into the dog, and you do not know whether the dog started before you noticed. Multiply a sixfold chemical variation by a guess and the product is not a measurement. It is a starting point for a conversation with a clinician.

This is exactly why the tool above shows you both published positions instead of returning a verdict. Our page on grapes and dogs goes further into the tartaric acid work if you want the longer version.

The timeline, which matters more than the count

Every source we read agrees that this poisoning is slow to declare itself and fast to become irreversible once it does.

Read that last row and the reason the sources sound urgent becomes obvious. The window in which this is treatable closes before the dog looks ill. Cornell puts the instruction plainly: “call your veterinarian or a pet poison control center immediately. Early action can significantly improve the chance of preventing or minimizing kidney damage.”

  hour 0    fruit eaten. dog looks normal.
     |      <-- this is the useful window
     |
  hour 6    vomiting or diarrhoea may start
     |      (MSD: 6-12 h)
     |
  hour 24   thirst, urination, weakness,
     |      limb swelling may appear
     |      (Cornell: 1-3 days)
     |
  hour 24-72  oliguric or anuric renal
     |        failure may develop (MSD)
     v
  once anuric: "most dogs die or are
  euthanized" (MSD)

  the phone call belongs at hour 0,
  not at the first symptom.

The things people ask that the sources do not answer

Does the type of grape matter? Nobody we read publishes a variety-by-variety breakdown. The tartaric acid range from ASPCA Pro implies it must matter, but no source turns that into guidance.

Are raisins worse than grapes by weight? Drying concentrates the fruit, and Cornell groups grapes, raisins and sultanas together as highly concentrated in the relevant compounds. We could not find a published multiplier, so we have not invented one, and the tool above treats a piece as a piece.

What about grape juice, wine, jam, or fruit cake? Not addressed by any source we fetched. Cream of tartar and tamarind are addressed, because that is where the discovery came from. If it is in your kitchen and it contains grapes, treat the question as one for the vet rather than for a web page.

Is my dog in the clear because it has eaten grapes before without harm? No source supports that reasoning, and the variable-concentration finding argues directly against it.

What we could not source

We could not find a published lower bound below which ingestion is considered inconsequential. MSD names a level above which risk is described; it does not name a level below which risk is excluded. Those are different statements and the difference is the whole page.

We could not find any published figure for cats. The MSD entry is titled for dogs. Cornell’s is a canine health centre. If a cat has eaten grapes, we have nothing sourced to tell you, which is itself worth knowing before you go looking.

We could not source the widely repeated claim that a specific number of grapes is lethal for a dog of a given weight. It is not in any of the three sources we read.

We could not source any at-home step that helps. There is nothing on this page about making a dog sick, because we are not qualified to write it and because the sources that discuss decontamination are addressed to clinicians.

Printable: what your vet will want in the first thirty seconds

Related pages on this site

If the food was something else, the poison control directory covers the kitchen line by line with its sources attached. For other published thresholds there are the onion and garlic checker, the xylitol checker, the chocolate calculator and the human painkiller checker. If it was a plant, use the poison control directory. If you are in the middle of it right now, go to my dog ate something or the triage helper, and before you call the vet will help you describe what you are seeing.

Cream of tartar, tamarind, and the other things on the same shelf

The discovery that reframed this poisoning came out of a batch of homemade playdough, which means the risk is not confined to the fruit bowl. ASPCA Pro states that "ingestion of cream of tartar or tamarind may lead to vomiting and acute kidney injury in dogs", and MSD now files tamarind alongside grapes and raisins in the title of its entry.

Cream of tartar sits in the baking cupboard. It appears in meringue, in some biscuit recipes, in snickerdoodles, and in the salt dough that gets made at kitchen tables every winter for tree decorations. Tamarind turns up in chutneys, in Worcestershire sauce, in pad thai and in some fizzy drinks. None of those is on a standard toxic-foods list, and none of the sources we read publishes a threshold for them.

We are not going to extrapolate one. What that finding does justify is a different question when you ring the vet. Instead of "he ate some grapes", the more useful opening is "he ate something containing tartaric acid, here is the packet". Bring the packet.

Why this checker no longer gives you a verdict

The earlier version of this tool sorted results into urgency bands and told you which one you were in. We have taken that out deliberately, and it is worth saying why on the page rather than quietly.

A band implies the tool knows something about your animal. It does not. It knows a number you typed, and that number rests on a count you made after the fact about fruit whose active compound varies by a factor of six. Two of the three sources behind this page decline to convert that into a threshold at all. A tool that converts it anyway is adding false precision to an emergency, and false precision in an emergency buys delay.

So the tool now does one job honestly. It puts your arithmetic next to what named sources published, tells you when each was read, shows you where those sources disagree, and then sends you to somebody who can actually examine the dog. If that feels like less than you wanted from a calculator, that gap is the accurate one.

⬇️ Download this data (CSV) - free to reuse with credit