Cycling a goldfish aquarium

A goldfish aquarium is a small waste-treatment plant, and cycling is the process of staffing it. The staff are bacteria, they take weeks to arrive in useful numbers, and until they do the fish is living in its own ammonia. Everything else about a new tank is secondary to this.

The nitrogen cycle in plain terms

Fish excrete ammonia continuously, mostly through the gills, and uneaten food and waste break down into more of it. Ammonia is toxic to fish at concentrations far too low to see or smell.

In an established tank, one group of bacteria living on the filter media takes up ammonia and converts it to nitrite. Nitrite is also toxic. A second group converts nitrite to nitrate, which is far less harmful and is removed by weekly water changes. Ammonia in, nitrate out, with the filter as the factory floor. A tank is cycled when both groups are numerous enough to process a day’s worth of the fish’s ammonia within that day, so the test kit reads zero for both even though the fish is producing both constantly.

The bacteria live on surfaces, overwhelmingly in the filter, and only a little in the water itself. This is why the filter is the thing you cycle, and why it must never be sterilised or thrown away.

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

What is actually growing

Worth two minutes, because every rule further down this page follows from it rather than having to be memorised separately.

Two groups of bacteria, not one. The first oxidises ammonia to nitrite; the second oxidises nitrite to nitrate. They are different organisms with different growth rates, which is why ammonia clears before nitrite does and why the nitrite stage is the long one.

They are slow. These are not the bacteria that turn milk sour in a day. They divide on a timescale of many hours, which is the entire reason cycling takes weeks and cannot be hurried by wanting it more — and why goldfish tank setup puts the waiting before the fish rather than after.

They live on surfaces, not in the water. Almost the whole colony is attached to media, glass, substrate and decor. That is why filter media volume matters more than pump flow, and why changing water does not remove the cycle.

They need oxygen. A filter that stops starves them, and the longer it stays off the more of the colony dies, which is why restarting one that has been off overnight is the dangerous part of a power cut rather than the outage itself — power cuts and emergencies covers it.

They need carbonate hardness, which they consume as a carbon source. Run out and the colony stalls and the pH falls, which is the crash covered below and in pH, KH and GH.

And drying kills much of the colony. Media kept wet in tank water carries a colony; media left on a draining board for an afternoon loses much of it, and the longer it dries the less survives.

Fishless cycling

This is the better route whenever you have the choice, because no fish is exposed to anything.

Set up the tank with the filter running and, if you have a heater, warm the water to the low-to-mid twenties Celsius (low seventies Fahrenheit) to speed things up. Then supply ammonia. The cleanest source is a bottle of pure ammonia solution, sold for cleaning, with no surfactants, perfumes or dyes; shake the bottle, and if it foams, it is the wrong one. The alternative is a pinch of fish food added daily, which rots into ammonia; it works but is slower and harder to control.

Dose ammonia to about 2 mg/L on the test kit, and no higher than 4 mg/L if your pH is below about 7.5, and watch. Over one to three weeks ammonia falls and nitrite appears and climbs; re-dose ammonia whenever it drops near zero so the first group of bacteria stays fed. Over the following two to five weeks nitrite peaks and falls as the second group catches up, and nitrate rises steadily. Do not overdose; very high ammonia stalls the process rather than hastening it, because the free ammonia that builds up at a higher pH inhibits the bacteria.

The cycle is complete when a dose of about 2 mg/L ammonia is processed to zero ammonia and zero nitrite within 24 hours, repeatably, with nitrate present. Then do a large water change to bring nitrate down and add the first fish within a day or two, because the bacteria begin to starve without a supply of ammonia. Expect the whole process to take 4–8 weeks, sometimes faster, occasionally slower.

What the test kit reads, week by week

Approximate, and the point is the shape rather than the dates. A seeded filter runs through the whole sequence in days; a cold tank can take twice as long.

Roughly	What the test kit reads	What is happening
Days 1–5	Ammonia high and steady, nitrite and nitrate zero	Nothing visible. The first bacteria are establishing and there are too few of them to register.
Week 1–2	Ammonia starting to fall, nitrite appearing	The first group is working. This is the first real evidence of progress.
Week 2–4	Ammonia clearing within a day, nitrite high and climbing, a trace of nitrate	The frustrating middle. The second group establishes far more slowly than the first.
Week 4–6	Ammonia zero, nitrite falling, nitrate rising	The nitrite plateau breaks, often quite suddenly.
Week 5–8	Ammonia and nitrite zero, nitrate clearly present	Cycled — once this repeats for several days on a full dose.

Two things to expect from that table. The nitrite plateau is the long part, it is normal, and almost everyone who abandons a cycle does it here. And nitrate appearing is the good news — it is the waste product of a filter that works, so a tank reading zero nitrate at week six has not finished.

Things that stall a fishless cycle

Low temperature, a pH crash, chlorinated top-up water, and ammonia dosed far above the target. If nothing has happened after three weeks, check those four first, and see the pH crash below for the one of the four you will not find unless you go looking.

Fish-in cycling

Many people arrive here because the fish came first. The tank can still be cycled with the fish in it; it simply takes more attention, and the fish take the strain.

Test ammonia and nitrite every day. Whenever either shows on the test at all, change 30–50% of the water with dechlorinated water at tank temperature, and change more if the reading is high. Feed lightly, once a day, only what is eaten within a minute, because every flake becomes ammonia. Do not add more fish, do not clean the filter, and do not medicate. Some water conditioners claim to bind ammonia temporarily; they may help at the margins but are no substitute for water changes.

This is the same process, spread thinner. Because you keep removing the bacteria’s food, it commonly takes as long as a fishless cycle or longer, and the water changes are the price of keeping the fish safe. Our guide to testing goldfish water covers reading the kit, and the water change guide the technique.

Seeding with mature media

The fastest way to cycle any filter is to borrow bacteria that already exist. A handful of sponge or ceramic media from a healthy, established aquarium, placed inside your own filter, can cut cycling from weeks to days. Keep the media wet and out of the cold on the way over, and take it only from a tank whose fish are clearly healthy, since it can carry pathogens as well as bacteria. Gravel, decor and a squeeze of the brown sludge from a mature sponge work too, less strongly. Judging whether a source is healthy enough to take from — and what to do when you genuinely cannot tell — is covered in buying second-hand, because a common way people acquire mature media is with somebody else’s filter.

Bottled bacteria

Bottled nitrifying bacteria range from genuinely helpful to inert, and results with the same product vary with its age and how it was stored. Treat a bottle as a possible head start rather than a substitute for the process: use it, keep dosing ammonia or testing daily, and let the test kit tell you whether it has done anything.

The pH crash, which is the invisible one

A common reason a cycle stops dead, and the only one on the list that you will not find unless you go looking.

Nitrification consumes carbonate hardness. Every time the bacteria convert ammonia they produce acid, and the KH in the water neutralises it. KH is therefore a consumable: it goes down over the weeks of a cycle and nothing replaces it unless you do. When it runs out, pH falls — often abruptly, from 7.5 to below 6.5 in a day or two — and nitrifying bacteria slow steeply below about pH 6.8 and are largely stalled by pH 6. The cycle appears to have failed for no reason.

It is most likely if your tap water is soft. The pH, KH and GH page covers how to read your own, and the source water page covers what comes out of your tap in the first place.

The fix is simple once you know to look. Test KH weekly during a cycle alongside ammonia and nitrite. If it is falling toward zero, a partial water change restores it, and a small amount of bicarbonate of soda raises it directly. Bring pH back up gradually rather than in one jump — a fast swing is its own stress if fish are present — and the cycle usually resumes within days of the water being habitable again.

Temperature and how long this takes

Nitrifying bacteria are temperature-sensitive to a degree that surprises people.

They work best in the low-to-mid twenties Celsius and slow steadily as the water cools; near 10 °C an established colony still works at a fraction of its warm rate, a new one grows very slowly, and the nitrite stage suffers most. A goldfish tank kept genuinely cool — which is otherwise good for the fish — will cycle noticeably more slowly than a tropical one, and a tank cycling in an unheated room in winter may take twice the usual time or appear to do nothing at all.

The practical move is to cycle warm and cool afterwards. If a heater is available, run the fishless cycle at around 24 C, then let the tank settle to the temperature the fish will actually live at before they arrive. The temperature guide covers where that should be. With fish already present, do not raise the temperature to hurry things along: warmer water holds less oxygen and makes ammonia more toxic, which is the wrong trade during the period when both are the problem.

New tank syndrome, and the cloudy water that is not the cycle

Two things worth naming because they cause a great deal of confusion in the first fortnight.

New tank syndrome is simply what an uncycled tank does to fish: ammonia and nitrite climbing while the filter is still empty, and fish that gasp, clamp their fins, sit on the bottom, lose their appetite and sometimes show red streaks in the fins. It has a name because it is so common, not because it is a distinct disease. The treatment is water changes and time, not medication — and medicating a tank in this state usually makes it worse by killing what bacteria there are.

The cloudy water is a different thing entirely. A white or grey haze that appears in the first week or two is a bacterial bloom: fast-growing heterotrophic bacteria feeding on dissolved organics, reproducing in hours rather than the days the nitrifying bacteria need. It is not the cycle starting, it tells you nothing about progress, and it clears by itself. The cloudy water page covers the other causes. Resist the urge to respond to it: changing the filter media or dosing a clarifier at this stage sets the real cycle back.

How to know it is done

Four things, together. Ammonia reads zero. Nitrite reads zero. Nitrate is present and rising. And this has held for several consecutive days while ammonia was being added, or with the fish present and no water changes needed. If the tank clears a 2 mg/L dose in a day, it is ready for its first fish; it is not yet ready for its full complement, which should be added gradually over the following weeks.

Mistakes that undo the cycle

• Rinsing filter media under the tap. Chlorine and chloramine can damage the bacteria. Rinse sponges in a bucket of water taken from the tank, and only when flow has dropped.
• Replacing cartridges on a schedule. The cartridge is the colony. Replace media only when it is falling apart, one piece at a time. If your filter is built around disposable cartridges, pack extra sponge or ceramic media around them so the bacteria have a permanent home; the filtration guide shows how.
• Letting the filter stop. A few hours without flow is survivable; a day is not, because the media goes anoxic and the colony dies back.
• Adding too many fish at once. The colony grows to match the load; it cannot jump ahead of it.
• Medicating without checking the label. Some treatments harm filter bacteria. Test the water before treating anything.

When an established tank loses its cycle

Not a beginner’s problem — this one catches experienced keepers, usually a year or two in, and it is easy to misread as disease.

The signal is ammonia or nitrite reappearing in a tank that has read zero for months. The usual causes:

• The filter stopped and stayed stopped. The media goes anaerobic within hours; the filtration guide covers why restarting it badly is worse than the outage, and power cuts covers the emergency itself.
• Media rinsed in tap water, or replaced all at once. A common self-inflicted version.
• A medication that kills bacteria. Many do, and some do not say so prominently; using medications safely is the page for that.
• Weeks with no feeding. The colony shrinks to match its food supply, so a long holiday leaves a filter sized for a fasting fish. The holiday guide covers how to avoid it.
• A big jump in stocking, which is not a crash so much as the colony being outgrown overnight.

Recovery is much faster than starting from scratch — days rather than weeks — because bacteria survive on the glass, the substrate, the decor and in the parts of the media that were not disturbed. Treat it exactly like a fish-in cycle: test daily, feed lightly, change water on the readings, add nothing, and do not clean anything until it reads zero again.

Cycling a quarantine or hospital tank

A specific case with a specific answer, because these tanks are always needed at short notice and never have weeks to spare.

You do not cycle them in advance. You keep a small sponge filter running permanently in the main tank, where it seeds itself for free, and move it across when a quarantine tank or a hospital tank is needed. The spare tank is then cycled from the moment it is filled.

This is the strongest practical argument for running two filters on the main tank at all, and it costs the price of a sponge and a share of an air pump. Without it, a hospital tank set up in an emergency is an uncycled tank containing an already-sick fish, which means daily water changes on top of whatever treatment is under way.

Why stocking gradually is not just caution

“Stock gradually” appears all over this site as an instruction, and it deserves its mechanism, because knowing it tells you how gradually.

The colony sizes itself to the supply. Bacteria grow to match the ammonia actually being produced and no further — there is no reserve capacity sitting idle. A filter cycled on a dose that supports two goldfish holds exactly enough bacteria for two goldfish.

So adding stock is adding ammonia to a colony that has not grown yet. Double the fish and you have roughly doubled the supply against a population that needs days to catch up. The result is a small, brief ammonia or nitrite reading while it does — a mini-cycle, which is harmless if you are watching and harmful if you are not.

Which gives the practical rule:

Adding	Wait before the next addition	Test
The first fish, after a fishless cycle	2 to 3 weeks	Ammonia and nitrite, every few days
Each fish after that	2 to 3 weeks	Same
A large fish, or several at once	3 to 4 weeks	Daily for the first week
After moving media to a new tank	2 weeks before adding anything	Daily at first

You will need a liquid test kit throughout, and strips are not adequate for the two readings that matter here — the test kit guide covers why, and the parameter checker will interpret what you get.

Quarantine makes this free. A new fish should spend at least four weeks in quarantine anyway, which is almost exactly the interval the filter needs — so doing both at once costs nothing and the two habits reinforce each other.

And the ceiling is still the ceiling. Adding gradually lets the colony keep up; it does not raise the number of fish the tank can hold, which how many goldfish you can keep sets.

Cycling, by situation

Pages across this site say “cycle it”, each in its own context. This is the consolidated version.

Situation	What to do
New tank, no fish yet	Fishless cycle with an ammonia source. Four to eight weeks
New tank, fish already in it	Fish-in cycle: daily testing, water changes on any reading. See below
Quarantine or hospital tank	Do not cycle it. Run a sponge filter in the main tank until needed
A second filter on a running tank	Nothing. It colonises from the tank over weeks
Moving house	Keep media wet in tank water. Expect a partial re-cycle either way
Upgrading the tank	Move the old filter across and run both for a month
A pond	Same sequence, easier for the volume, slower in cold water
After the filter ran dry or was cleaned in tap water	Treat it as a new cycle and test daily
After losing a fish for no clear reason	Test first — an uncycled or crashed filter is a common answer

The one that surprises people is the quarantine row. A quarantine tank is not cycled and does not need to be, because it runs on a sponge filter that has been colonising in the display tank and on daily water changes — the hospital tank page covers why that combination works and a cycled quarantine tank standing empty does not.

After the cycle

Stock gradually, change water weekly, and keep testing ammonia and nitrite once a week for the first couple of months. A cycled filter is the foundation that the rest of goldfish care stands on, and it is far easier to protect than to rebuild.

What a settled tank reads like, which nobody tends to say out loud after weeks of watching numbers move: ammonia zero, nitrite zero, and nitrate rising at a rate you can predict. That third one is the useful reading now. Take a nitrate test right after a water change and another just before the next, and the gap between them is your tank’s slope. Once you know it — say it climbs from 5 to 20 mg/L over a week — you have a baseline that tells you more than any single reading ever will, because a week where it climbs to 50 instead means something changed before any fish looks unwell.

Then taper the testing, rather than stopping it. Most people go from daily to never, usually in one step, and the kit is dry the day it matters. A workable schedule is every few days for the first month with fish, then the settled cadence the testing guide sets out: nitrate and pH weekly, ammonia and nitrite every few weeks. The weekly nitrate reading is the one that earns its place here, because it is the slope rather than the number that carries the information. The water parameter checker will read a set of results if the numbers stop meaning much to you.

Go back to daily for a week whenever the system changes — a fish added, a filter replaced or cleaned, a heater failed, a course of medication, a holiday where someone else fed. Those are the moments the colony is asked to do something new, and they are the only times frequent testing earns its keep. The sections above on what undoes a cycle cover what to do if one of them goes wrong.