Testing goldfish water: ammonia, nitrite, nitrate and what the numbers mean

A goldfish cannot tell you the ammonia is climbing, and by the time its behaviour does, damage has started. Testing is how you find out a week earlier. It is also how you learn whether your routine is working, whether the tap water is helping or hurting, and whether a sick fish is sick because of the water, which it usually is. Kits are cheap, the readings take ten minutes, and the habit prevents most of the emergencies that fill goldfish forums.

Liquid kits versus strips

Test strips are dipped in the tank, then compared against a colour chart after a set time. They are quick and cost little to start with, but the colours are hard to read at the low end, they degrade in humid air once the tube is open, and some strips do not test ammonia at all. They are useful as a fast weekly screen once a tank is stable.

Liquid kits use reagent drops in a test tube and are read against a chart after a few minutes. They are more precise, especially for the two readings where precision matters most, ammonia and nitrite, and they work out cheaper per test over a year. A master kit covering ammonia, nitrite, nitrate and pH is the standard recommendation for a goldfish tank, with a KH test added if your tap water is soft. Our buying guide to aquarium test kits compares specific products; it is a shopping guide, and nothing on this page depends on which kit you choose.

What to test, and the target ranges

Parameter	Target for goldfish	Act when
Ammonia	0 mg/L	Any reading above zero
Nitrite	0 mg/L	Any reading above zero
Nitrate	Ideally under 20 mg/L, comfortably under 40	Climbing toward 40 mg/L
pH	7.0 to 8.0 ideal; 6.5 to 8.5 tolerable	Outside the range, or shifting by more than about 0.5 in a day
KH (carbonate hardness)	Roughly 4 dKH (about 70 mg/L) or higher	Below about 3 dKH, when pH becomes unstable
GH (general hardness)	Wide tolerance; moderate to hard water is fine	Rarely a problem for goldfish
Temperature	18 to 24 °C for fancies; 10 to 24 °C for single-tails	Swings of several degrees in a day

Ammonia and nitrite are the emergency readings. Any measurable amount means the biological filter is not keeping up, whether because the tank is new, the filter was over-cleaned, a fish died unnoticed, or stocking has outgrown the setup. Nitrate is the maintenance reading: it rises steadily between water changes, and how fast it rises tells you whether your water change schedule fits your tank. pH is the stability reading, and the least worth chasing; goldfish tolerate a broad range as long as it stays put.

How often to test

While a tank is cycling, test ammonia and nitrite every day and change water whenever either shows, as the cycling guide describes. Keep the daily habit for the first week or two after adding a new fish or replacing filter media.

Once a tank is stable, weekly is enough: nitrate and pH every week, ammonia and nitrite every few weeks or whenever something changes. Test immediately, before doing anything else, if a fish is gasping at the surface, sitting on the bottom, refusing food, or showing red streaks or clamped fins. Water quality is behind most of these signs, and the test tells you whether the fix is a water change or something more.

Between the tests, and the badge that covers it

Everything above is a spot check, and a spot check tells you about one moment. Ammonia rarely fails on a schedule: a filter stops overnight, a fish dies behind the decor on a Friday, a medication wipes the colony out, and by the next weekly test the reading is back to zero with the damage already done.

A continuous in-tank ammonia sensor covers that gap. It is a small badge stuck inside the glass with a pad that changes colour in the presence of free ammonia, needs no sample, no reagents and no timing, and is read by glancing at it whenever you pass. Ours are listed in the test kit buying guide.

What it is good at is exactly what a weekly kit is bad at: telling you that something changed in the days between tests. Because goldfish tanks fail on ammonia more than on anything else, a device that watches it constantly is worth more here than in most aquariums.

What it is not is a replacement for a liquid kit. It reads only free ammonia, not the total figure a liquid test gives, and the two diverge with pH and temperature. It gives a band rather than a number, so it cannot tell you whether a reading is drifting. And the pad expires, typically after several months to a year, degrading to a permanently reassuring colour rather than an obviously dead one — which is the failure mode to know about. Write the replacement date on the glass beside it.

Treat it as a smoke alarm rather than a thermometer: it is there to make you pick up the test kit, not to save you from owning one.

Taking a sample worth testing

Small thing, and it changes results more than people expect.

• Take it from mid-water, away from the filter return. The water leaving the outflow is the best-processed in the tank; the water in a dead corner is the worst. Mid-tank, mid-depth is the true average.
• Use a clean container, rinsed in tank water and never in soap. A jug kept for the purpose is easiest.
• Test before a water change, not after. After a change you are measuring the tap water with some tank water in it, which tells you nothing about how the tank is doing between changes. Nitrate in particular is only meaningful at its weekly peak.
• Do not test in the hour after feeding, or immediately after disturbing the substrate, both of which put a brief spike into the water column.

Reading the colours

Read tubes in daylight or under a white light, against the white part of the chart, at the time the instructions specify; readings drift if left. The hardest judgment is the low end of the ammonia and nitrite scales, where zero and the first step look alike: if in doubt, treat a maybe as a yes and change some water.

A few practical points: some nitrate reagents need vigorous shaking for a full minute or the result reads low; reagents have expiry dates and go wrong after them; and straight after a water change a reading may be real ammonia from chloramine, or ammonia that a conditioner has bound into a less toxic form which some kits, Nessler-type ones especially, still count. Retest after a day and treat it as real until then. Testing before the change rather than after avoids the confusion.

What the pattern means

The nitrogen cycle

1. 1
2. 2
3. 3
4. 4

1. 01 Source Fish waste Gills, droppings, uneaten food, decay Most of the ammonia is excreted by the fish across the gills; the solids break down into more. becomes ammonia
2. 02 Toxic Ammonia NH₃ / NH₄⁺ Damages the gills and, at higher levels, the nervous system; clear water can still be toxic. filter microbes turn it into nitrite
3. 03 Toxic Nitrite NO₂⁻ Stops the blood carrying oxygen properly. filter bacteria turn it into nitrate
4. 04 Far less harmful Nitrate NO₃⁻ Removed by water changes and used by plants.

water changes take the nitrate out, and the cycle starts again

Individual numbers are less informative than the shape they make together. This is the table to read with your results in front of you.

What you read	What it usually means
Ammonia high, nitrite and nitrate zero	A new tank at the start of its cycle — or an established filter that has just been destroyed, by a tap rinse, a media replacement, or being left off too long.
Ammonia falling, nitrite high, a trace of nitrate	The middle of a cycle. The first group of bacteria is working and the second is still catching up. Normal, and the longest part.
Ammonia and nitrite zero, nitrate rising between changes	A working tank. This is the pattern you want, and the rising nitrate is the evidence rather than a fault.
All three zero, on an established stocked tank	The filter is probably not converting anything. Suspicious rather than excellent.
Ammonia and nitrite zero, nitrate very high	The filter is fine and the water changes are overdue.
A trace of ammonia, nitrite zero, nitrate high	An established tank that has just been overloaded: a fish died unnoticed, the filter was over-cleaned, stocking jumped, or feeding did.
Ammonia and nitrite both present, nitrate high	Several things at once, usually a neglected tank. Daily water changes, and feed nothing.

Two readings deserve separate mention because they are counter-intuitive.

Nitrate at zero on a months-old stocked tank looks like the cleanest possible result and is usually a warning. Nitrate is what a working filter produces; its absence means the conversion is not happening. Cycling a goldfish aquarium covers what to do.

A trace of ammonia with everything else normal is a common real finding, and it means the filter is slightly behind rather than absent. Change water, feed lightly, test again tomorrow, and look for what changed.

When the readings lie

Most surprising results are the test being wrong rather than the tank, and the list of reasons is short enough to check.

• Conditioner and chloramine readings. Dechlorinating chloramine releases ammonia, and some conditioners bind that ammonia into a less toxic form which certain kits, Nessler-type ones especially, still count. A reading just after a water change can therefore be real ammonia or the kit. Retest after a day and treat it as real until then. Test before a water change rather than after, and see red streaks for what that means mid-emergency.
• Expired or exhausted reagents. They have dates and they go wrong afterwards, usually reading low. If a kit has been open for years, replace it before trusting a surprising result.
• The nitrate bottle that needs shaking. Many nitrate reagents require vigorous shaking — of the second bottle in particular — for a full minute, or they read far lower than the truth. This is a common false reassurance in fishkeeping.
• Tubes that were not rinsed, or were rinsed with tap water, or still carry reagent from last week.
• Reading at the wrong moment. Colours develop and then drift; read at the time the instructions state, not when you get back to it.
• Reading in the wrong light. Daylight or a white bulb, looking down through the tube against the white part of the chart, not sideways under a warm-toned lamp.
• Strips that have been open in a humid room. They degrade quickly once the tube is unsealed and the desiccant is spent.

The habit that catches most of these: when a result surprises you, run it again before acting on it, ideally with a fresh sample.

When the fish and the numbers disagree

Worth thinking about clearly, because both directions happen and they mean different things.

Clean water, unwell fish. The readings have ruled out common causes, which is real information and not a dead end. What they do not rule out: parasites, an injury, an internal infection, a temperature swing that has already passed, gill damage from an earlier problem that is still healing, or something no kit measures. Look next at temperature and its recent history, at anything added in the last fortnight, and at the fish itself under a torch — the health section is organised by symptom for this.

Bad water, fish that look fine. Believe the numbers. Goldfish are undemonstrative about chronic problems and will behave normally through conditions that are doing steady damage, which is the whole reason for testing in the first place. A fish looking well is not evidence that a reading is wrong.

And the reading that changed without you. A parameter that has moved with no obvious cause is worth repeating and then investigating rather than accepting: a dead fish behind the decor, a hand cream in the tank, a heater that failed, a filter that stopped overnight.

Testing a pond

The same parameters, a few differences in how you take the sample.

• Sample at depth and away from the inflow, not from the surface next to the filter return, which is the best-oxygenated and most recently processed water in the pond.
• Sample at the same time of day. pH in a planted pond rises through the day as plants consume carbon dioxide and falls overnight, and a swing of half a unit or more between dawn and dusk is normal rather than alarming.
• Test after events rather than on a schedule: heavy rain, an algae die-off, a hot week, new fish, or the spring restart — a pond through the year covers which events matter.
• Expect nitrate to behave differently. A planted pond can hold nitrate low without water changes, which is one of the genuine advantages outdoors, and goldfish ponds covers what that does and does not excuse.

What not to bother testing

Saving money and shelf space, because a great deal is sold that a goldfish keeper does not need.

• Chlorine. Handled by dosing a conditioner every time, which is cheaper and more reliable than measuring. The source water page covers the chloramine distinction that does matter.
• Phosphate. Relevant to algae management in high-tech planted tanks. In a goldfish tank, nitrate and light tell you what you need to know, and algae covers the rest.
• Copper, unless you have old plumbing you are suspicious of, or have used a copper-based treatment and want to confirm it has gone.
• Dissolved solids meters. Interesting, cheap, and rarely actionable for goldfish. The single exception is diagnosing a tank that has been topped up rather than changed, where minerals accumulate and TDS climbs while every other reading stays normal; the source water page covers that, and a bigger water change settles it without buying anything.
• GH, beyond knowing it once. Goldfish are unfussy about general hardness.

A good liquid master kit plus a KH test if your water is soft covers essentially everything, and the money saved is better spent on the tank.

KH and GH, briefly

Both get a full treatment in pH, KH and GH, including the pH crash and why the same ammonia reading is more dangerous in alkaline water. The short version:

KH, or carbonate hardness, is the water’s buffering capacity: its ability to resist a fall in pH. The bacteria that process ammonia produce acid as they work, and a goldfish tank produces a lot of ammonia. In water with a healthy KH that acid is absorbed and the pH holds; in soft water the KH is used up between water changes and the pH can drop sharply, sometimes to a level where the filter bacteria themselves stall. That is the pH crash, and it explains why some tanks read fine on Monday and are in trouble by Friday. If your tap water has a KH below about 3 to 4 dKH, a small bag of crushed coral or aragonite in the filter is a common way to add buffering slowly.

GH, or general hardness, measures dissolved calcium and magnesium. Goldfish are unfussy about it and do well in moderately hard water; it is worth knowing but rarely worth adjusting.

Test your tap water

Do this once, and again if your water company changes its source. Test pH, KH, nitrate and, if your kit offers it, ammonia, on water straight from the tap and again on a glass left to stand for 24 hours; dissolved gases can make fresh tap water read a different pH from the settled value your tank will actually sit at. Tap nitrate of 10 to 20 mg/L or more is not unusual in farming regions and changes how much a water change can achieve. Knowing your baseline turns every later reading into a comparison rather than a mystery, and the water change guide explains the options when tap water is part of the problem.

Keep a log

Write the readings down, with the date and anything you changed: fish added, food changed, filter rinsed, water changed. A notebook by the tank or a note on your phone is enough. Single readings tell you where you are; the log tells you which way you are heading, and a nitrate line that climbs a little faster each month is the earliest warning of overstocking you will get. The water parameter checker will interpret a set of readings for you; the log is what turns them into a trend.