A test kit gives you a number. Almost nothing tells you what the number means, and for ammonia the reading on its own does not tell you whether the fish are in danger. This page goes through each test, says what the reading is measuring, and gives the published figure it should be read against, with the document it came from. Where the authorities give different answers, both are here.
Ammonia in water exists in two forms. One of them, ammonium, is relatively harmless. The other, un-ionized ammonia, is the one that damages fish. A liquid or strip test kit measures both together and reports the total, so the same reading can mean almost nothing or mean an emergency, depending on the pH and the temperature of the water.
Both are pushed by the same chemistry. Higher pH and higher temperature move more of the total into the toxic form. The University of Florida's extension service publishes the conversion table and puts the damage threshold plainly:
Anytime the UIA is higher than 0.05 mg/L, damage to fish tissues can occur.
The Southern Regional Aquaculture Center, run out of Mississippi State, gives a slightly lower figure for long exposure: chronic damage to gills and kidneys "as low as 0.06 mg/L", and roughly 0.6 mg/L as acutely lethal. It also publishes the full conversion table, and one line from a companion sheet is the single most useful thing an aquarium owner can know: "less than 10% of ammonia is in the toxic form when pH is less than 8.0."
Here is what that does to one reading. The numbers are the toxic un-ionized fraction of a total ammonia reading of 1.0 mg/L, at 26 degrees C (79 F), calculated with the formula Florida's environmental agency publishes and checked against the printed table:
| pH of the water | Toxic ammonia, mg/L | Against the 0.05 mg/L damage figure |
|---|---|---|
| 6.5 | 0.002 | Well under it. |
| 7.0 | 0.006 | Well under it. |
| 7.5 | 0.019 | Under it. |
| 8.0 | 0.057 | Over it. |
| 8.5 | 0.161 | Three times over it. |
| 9.0 | 0.378 | Approaching the figure quoted as acutely lethal. |
The same bottle of water, the same test result, and at one end of that table the fish are fine while at the other they are being burned. That is why a shop can look at your test strip and tell you the reading is fine without asking what your pH is, and be wrong.
Put your own reading through the calculator. It takes your ammonia result, your pH and your temperature and gives you the toxic figure.
Nitrite is what ammonia becomes on its way to nitrate. It poisons fish in a specific way: it crosses the gill, binds to hemoglobin and turns it into methemoglobin, which cannot carry oxygen. The blood goes brown, which is where the pond name comes from. A fish can suffocate in water with plenty of oxygen in it.
Chloride blocks it, because chloride and nitrite compete to cross the gill. Florida's extension service puts the mechanism as the two "particles compete for space to cross the gills". This is the reason salt is used in fish emergencies, and it is real. What is not settled is how much.
| Who | Chloride to nitrite ratio | Minimum chloride they ask for |
|---|---|---|
| UF/IFAS Extension, Florida | 6 to 1 | 20 ppm |
| Auburn University Extension | 9 to 1, and notes many use 10 to 1 | 60 to 150 ppm |
| Southern Regional Aquaculture Center | 10 to 1 | 100 ppm |
| Merck Veterinary Manual | States the mechanism, publishes no ratio | Not given |
One thing is worth saying plainly about all three numbers. They come from channel catfish in ponds, because that is who paid for the research. Nobody has published a figure for tetras, cichlids or koi. If you use salt in an aquarium emergency you are borrowing a catfish number, and the honest position is that the mechanism is established and the dose is not.
Nitrate is the end of the chain and the least dangerous of the three, which is why water changes exist: the filter turns ammonia into nitrite into nitrate, and nothing in the tank removes the nitrate for you.
Ask what level is safe and four organizations give four answers.
| Who | What they publish |
|---|---|
| Merck Veterinary Manual | Under 20 mg/L |
| OATA, the UK aquatic trade body | No more than 50 mg/L above the level in the tap water |
| OATA, on its own catfish care sheet | No more than 20 mg/L above the tap water |
| RSPCA Australia | Under 50 ppm |
The US Environmental Protection Agency publishes national water quality criteria for a long list of substances, and it has no national recommended aquatic life criterion for nitrate, or for nitrite either. So the numbers above are guidance figures, not toxicology. Treat rising nitrate as a sign that the water changes are not keeping up, rather than as a line you cross at 21 and not at 19.
These get muddled constantly, and the extension literature says so out loud: "Hardness is commonly confused with alkalinity."
pH is how acidic the water is right now. It is not a fixed property of a tank. It falls overnight and rises during the day, because plants and algae take carbon dioxide out of the water in the light and everything in the tank puts it back in the dark. The pond sheet describes it as pH falling through the night "as respiring organisms add carbon dioxide to the water". If you test at different times of day you will get different answers, and neither is wrong.
Alkalinity, sold as KH, is the water's ability to resist that swing. It is the buffer. Low alkalinity is what turns a normal daily cycle into a crash, and the same sheet says that in waters with low alkalinity pH "can reach dangerously low ... or dangerously high ... levels". Extension guidance puts the minimum at about 20 mg/L as calcium carbonate, and the comfortable range at 75 to 200.
Hardness, sold as GH, is the dissolved calcium and magnesium. It is not the buffer, and changing one does not necessarily change the other. The US Geological Survey classifies it: under 60 mg/L is soft, 61 to 120 moderately hard, 121 to 180 hard, over 180 very hard. Your water company publishes your figure.
The failure this causes has a name. Nitrification produces acid and consumes alkalinity, so a tank that is never water-changed slowly runs its buffer down, and then the pH does not drift, it falls off a cliff. Merck calls it old tank syndrome and describes "a sudden and drastic drop in pH, often below 6.0, which kills bacteria in the biofilter." The fish die of the ammonia that follows, not of the pH.
Warm water holds less oxygen. This is not a small effect. Florida's extension service gives the saturation figures: water at 90 degrees F "can only hold 7.4 mg/L DO at saturation, whereas water that is 45 degrees F can hold 11.9 mg/L". The same service recommends 5 mg/L or more for fish health, and the US Geological Survey defines water under 2 mg/L as hypoxic.
So a heatwave, a heater stuck on, or a summer room does two things at once: it lowers the oxygen in the water and raises the fish's demand for it. Fish at the surface gulping in hot weather are not hungry.
This is the clearest published table, from the Merck Veterinary Manual, for fresh water.
| Test | Merck's reference range |
|---|---|
| pH | 6.5 to 9.0 |
| Total ammonia | 0 |
| Nitrite | 0 |
| Nitrate | Under 20 mg/L |
| Alkalinity | Over 100 mg/L as calcium carbonate |
| Hardness | Over 20 mg/L |
| Dissolved oxygen | Over 5 mg/L |
| Chlorine | 0 |
Species ranges narrow that. RSPCA Australia publishes 20 to 24 degrees C (68 to 75 F) and pH 6.5 to 7.5 for goldfish, with a minimum of 50 liters, about 13 gallons, and 24 to 26 degrees C (75 to 79 F) for tropical fish with ammonia under 0.1 ppm and nitrite under 0.2 ppm. OATA's care sheets give temperature, pH, GH and KH per group of fish rather than per species.
OATA's own testing sheet says ammonia and nitrite "should always be zero" and asks for testing "at least once a week".
There is also a legal answer, which surprises most people. In England and Wales the Animal Welfare Act 2006 defines an animal as "a vertebrate other than man", which puts fish inside it, and requires that an animal's needs are met including "its need for a suitable environment". The government's statutory guidance for anyone selling pets goes further and makes it a record-keeping duty: "water quality must be tested and recorded. The frequency of testing must be appropriate to the system, but must be at least weekly."
The same document is also the best answer to the oldest rule in the hobby. On how many fish a tank can hold, it says: "Numbers of fish should not be specified as they are determined by the water quality in the aquarium."
We went looking for the source of this rule. It is on hundreds of websites. It is not published by any extension service, any veterinary reference, any regulator or any trade body that we could find. The only official statement on stocking we found says the opposite: that numbers should not be specified, because the water decides. If someone tells you a tank holds ten inches of fish, ask them who says so.