“Four to six times turnover per hour.” It is the first thing anyone tells a new fishkeeper, it is printed on half the boxes in the shop, and taken literally it will send you home with the wrong filter.
Not because the ratio is bad. Because aquarium filter GPH and turnover get quoted in a unit your tank will never reproduce. Manufacturers measure flow from a bare pump, with no media in the chamber, no hoses and no head height. Assume real output is 30 to 50 percent lower once the filter is loaded and installed, which means a 29 gallon tank targeting 150 GPH of genuine throughput needs a filter rated somewhere around 200 to 250.
The rule of thumb also comes from an era when hang-on-back filters dominated and everyone kept broadly similar community tanks. It survives because it is roughly right for that case. Goldfish and larger cichlids want six to ten times. Bettas, shrimp and fry want two to four. Heavily planted tanks can run lower, because the plants take up ammonium directly.
Here is the thing nobody puts on the box: a filter’s job is not to move water. It is to hold a large, oxygenated colony of nitrifying bacteria and to pass tank water through that colony often enough. Flow serves that. It is not the goal.
So the useful question is not “how many GPH”, it is “how much real throughput, through how much usable media, for the bioload I actually have”. This post gets you to a defensible answer for your specific tank, with the arithmetic shown.
The short answer
Aim for four to six times your tank’s real water volume per hour, measured as actual flow rather than rated flow. A 29 gallon (110 litre) community tank wants roughly 120 to 180 GPH going through the media.
Because rated GPH is measured with no media, no hoses and no vertical lift, assume real output is 30 to 50 percent lower once the filter is loaded and installed. That means a 29 gallon tank targeting 150 GPH real needs a filter rated somewhere around 200 to 250 GPH.
Adjust for stocking rather than for tank size alone. Goldfish and larger cichlids want six to ten times. Bettas, shrimp and fry want two to four. Heavily planted tanks can run lower because the plants take up ammonium directly.
More media capacity is always good. More current is not, and neither is more fish than the volume supports, which the stocking calculator will tell you in about a minute.
What turnover actually means
Turnover is a ratio, not a rate. Divide the filter’s flow in gallons per hour by the tank’s volume in gallons and you get “times per hour”. A 200 GPH filter on a 20 gallon tank is 10x. The same filter on a 55 gallon tank is 3.6x.
The 4x to 6x figure is a rule of thumb from an era when hang-on-back filters dominated and everyone kept broadly similar community tanks. It survives because it is roughly right for that case. It breaks down in three situations, and it is worth knowing which one you are in.
It breaks down for messy, heavy-waste fish. A trio of fancy goldfish (Carassius auratus) in a 40 gallon (150 litre) produce a volume of waste that a community tank of tetras does not approach. Six to ten times turnover, plus generous mechanical media, is closer to right.
It breaks down for low-flow species. Betta splendens comes from slow, vegetated water, and a long-finned show betta swims in still water like a person walks in a wedding dress. Two to four times is plenty.
And it breaks down for planted tanks, in both directions. Plants strip ammonium, which reduces the biological load on the filter. Aggressive flow, meanwhile, drives off dissolved carbon dioxide, which many planted tank keepers are actively trying to keep in the water. So a heavily planted low-tech tank might run at 3x quite happily, while a high-tech injected tank often runs higher flow deliberately for nutrient distribution. Real disagreement exists here, and I would not pretend otherwise.
Why the number on the box is not the number in your tank
Manufacturers quote flow from the bare pump, with no media in the chamber, no hoses, and no head height. It is a fair measurement of the pump. It is not a measurement of your filter.
Four things eat into it.
Head height comes first. Every pump has a curve where output falls as it lifts water higher, and a canister sitting on the floor under a tank stand is lifting perhaps 3 feet (about 90 cm). Hang-on-back filters lift much less, maybe 6 inches (15 cm), which is one reason their rated numbers hold up slightly better in practice.
Media resistance comes second, and it is the big one. Push water through two foam blocks and a litre of ceramic rings and you have added serious back pressure. A canister with a full media basket routinely delivers well under its rated figure from day one.
Hose length, bends and intake strainers add more. A canister with 5 feet of ribbed hose and two 90-degree turns flows noticeably less than the same unit on a short straight run.
Then there is dirt. A filter that starts at 70 percent of rated flow will drop further as the mechanical stage clogs, which is precisely what it is meant to do. That decline is normal, and it is why the maintenance interval matters more than the specification.
Add it up and 30 to 50 percent below rated is a sensible planning assumption for a canister, with hang-on-backs at the gentler end of that range. I have never seen a manufacturer publish this, and I would be delighted to be proven wrong.
Biological capacity is media, not flow
Two identical filters, one packed with ceramic rings and one with a single thin cartridge, will not support the same bioload even at identical flow. The bacteria live on surfaces. Surface area is the currency.
This is where marketing gets slippery. Media packaging quotes enormous surface areas, sometimes hundreds of square metres per litre, and those figures typically come from techniques that measure every micropore in the material. Nitrifying bacteria are around 1 micrometre across and need oxygenated water flowing past them, so a large share of that quoted area is unreachable. Treat the numbers as a rough ranking between products rather than an absolute.
What actually matters, in rough order: total volume of media in the filter, how open its structure is so water flows through rather than around, and how well oxygenated that water is. A big sponge in a well-fed filter outperforms a small quantity of expensive sintered glass most of the time.
There is also a comfortable margin built into how filters behave. Bacterial colonies size themselves to the food supply, so an oversized filter does not maintain an oversized colony forever. What it gives you is headroom: the ability to absorb a slug of ammonia when a fish dies behind the driftwood, or when you add six new corydoras at once. That headroom is the actual argument for buying bigger than you need.
The four types, honestly compared
| Type | Typical real flow | Media capacity | Best for | Main weakness |
|---|---|---|---|---|
| Sponge (air-driven) | Low, hard to quantify | Good for its size | Fry, shrimp, bettas, hospital and quarantine tanks | Poor fine mechanical filtration, visible in the tank |
| Internal (powerhead) | Moderate | Small | Small tanks, turtles, temporary setups | Takes up tank space, limited media |
| Hang-on-back | 60-80% of rated | Moderate, expandable if you ditch cartridges | Community tanks to roughly 55 gal / 210 L | Surface skimming can trickle noisily, evaporation |
| Canister | 50-70% of rated | Large | Tanks over 40 gal / 150 L, heavy stocking, planted | Cost, 30-minute servicing, priming faff |
A word on hang-on-backs, since they are what most people buy. The models with a media basket you can fill yourself, rather than a slot for a proprietary cartridge, are worth the extra money several times over. Cartridge-based designs are engineered around a consumable, and every cartridge change throws away a chunk of your bacterial colony.
Sponge filters deserve more credit than they get. Two large dual sponges on a decent air pump will biologically outperform a mid-range hang-on-back on many tanks, cost a fraction as much, cannot trap a fry, and never fail catastrophically. They just look like sponge filters, which is why people avoid them in display tanks.
Media order, and where the hobby disagrees
Water should hit mechanical media first. Coarse foam or floss catches the physical debris so it never reaches the biological stage. Skip this and your ceramic rings pack with detritus, flow drops, and the media you paid for stops doing its job.
Then chemical, if you use any. Activated carbon has real uses (removing medication after treatment, clearing tannins, pulling odours) and no particular use as a permanent resident. It exhausts within weeks and then sits there as expensive biological media. Purigen and similar resins are a different case with genuine ongoing value in heavily stocked tanks.
Biological last. Ceramic rings, sintered glass, coarse open foam, plastic bio-balls above waterline in a sump. The water reaching this stage should be physically clean so the media stays open.
The disagreement is about polishing floss. One camp puts fine floss at the very end, after biological, so the last thing water passes through is the finest mechanical stage and the tank looks glass-clear. Another camp puts fine floss before biological, arguing the bio media should never see fine particulate at all. Both work. I lean towards fine floss last for water clarity, then a coarse pre-filter sponge on the intake so the heavy stuff never enters the canister in the first place. A pre-filter sponge is a two-pound upgrade that changes how often you have to open the filter.
Worked example: sizing a filter for a 29 gallon community
Say Yasmin has a 29 gallon tank, which her local shop sells as 110 litres. She plans eight harlequin rasboras (Trigonostigma heteromorpha), six peppered corydoras (Corydoras paleatus) and one bristlenose pleco (Ancistrus sp.).
Step one, real volume. Two inches of sand, a piece of spiderwood and two rocks displace a fair amount, and the tank is never filled to the rim. Call it 25 gallons (95 litres) of actual water. Everything from here uses 25, not 29.
Step two, target turnover. It is a community tank with a pleco, which is a messy fish, so she picks the upper end of the standard band: 6x.
25 gallons × 6 = 150 GPH of real flow.
Step three, convert to a rated figure. She is considering a hang-on-back, so she assumes 30 percent loss.
150 ÷ 0.70 = 214 GPH rated.
Step four, shop. An AquaClear 50 is rated at 200 GPH with a media basket she can fill herself. That is just under her calculated figure, which is fine because these estimates carry more uncertainty than 7 percent. A Fluval 207 canister sits at roughly 200 GPH rated, but at canister loss rates (say 40 percent) that gives about 120 GPH real, below her target, so she would want the next size up.
Step five, decide. She takes the AquaClear 50, fills the basket with coarse foam at the bottom and ceramic media above, fits a pre-filter sponge on the intake, and adds a small sponge filter in the far corner. Combined, that gives her comfortable capacity, redundancy if the hang-on-back fails, and a seeded sponge she can move to a quarantine tank later.
If your tank came in litres and the filters are sold in GPH, or the reverse, the unit converter handles the swap without a decimal slip. And working out “what percentage of rated flow am I really getting” is one line in the percentage calculator.
Sizing your filter, step by step
- Measure or estimate real water volume. Take the stated tank size and subtract roughly 10 to 15 percent for substrate, hardscape and the gap at the top. A 40 gallon (150 litre) tank with a deep planted substrate might hold 33 gallons (125 litres).
- Pick a turnover multiplier from your stocking, not your tank. Community 4-6x, messy or heavily stocked 6-10x, low-flow species 2-4x, planted low-tech 3-5x.
- Multiply. Volume × multiplier = target real GPH.
- Divide by your expected efficiency. 0.7 for a hang-on-back, 0.55 to 0.6 for a canister on a normal stand height. That gives target rated GPH.
- Shortlist filters at or above that rated figure, then ignore the flow number and compare media volume. Litres of media capacity is the spec worth optimising.
- Check adjustability. A flow valve or an adjustable outflow matters more than you expect, because it lets you buy for capacity and run at whatever current your fish tolerate.
- Plan the media stack before you buy, and budget for it. Many filters ship with cartridges you will replace on day one.
- Add a backup path. A cheap sponge filter running alongside is the difference between a dead pump being an annoyance and a dead pump being a fish loss.
Before you finalise any of that, it helps to know what bioload you are actually filtering. The Aquarium Stocking Calculator works from real water volume and species-specific waste output rather than the inch-per-gallon shorthand, so you can see whether your planned stock list sits comfortably inside your tank or quietly over it. It runs in your browser, nothing gets uploaded, and there is no account to create. Sizing the filter to a stocking plan you have actually checked beats sizing it to a number on a box.
The flow problem, and how to fix it
Some fish cannot cope with the current a correctly sized filter produces. Bettas are the obvious case, but the list includes gouramis, most long-finned fancy varieties, sparkling gouramis (Trichopsis pumila), many killifish and Endler’s livebearers in small tanks.
The signs are clear once you know them. A fish holding position behind a rock permanently. A betta resting on a leaf or the substrate more than it swims. Fins pinned. Food swept past before anyone can eat it. Substrate scoured into a bare patch under the outflow.
Fixes, roughly in order of how much I like them:
- Turn the flow valve down, if the filter has one. Free, instant, reversible.
- Aim the outflow at the glass or along the back wall rather than down the length of the tank. Flow that hits a surface dissipates.
- Fit a spray bar and point it at the rear glass or slightly upwards. This spreads the same volume over a wider area at lower velocity.
- Wedge a piece of coarse filter sponge over the outflow lip of a hang-on-back. It diffuses the waterfall into a broad trickle.
- Cut the bottom off a small plastic bottle and clip it over the outflow as a deflector. Ugly, effective, costs nothing.
- Add floating plants. Salvinia, frogbit and duckweed break surface movement and give the fish cover, which changes behaviour as much as the flow reduction does.
What I would not do is buy a smaller filter. You want the media capacity. Reduce the current instead, and keep enough surface agitation for gas exchange, because a completely still surface is its own problem. If you are stocking a betta community, the compatibility side is covered in betta tank mates that actually work.
Where people get caught out
Rinsing biological media in tap water is the classic. Chlorine and chloramine kill nitrifying bacteria on contact, and people do this in good faith because the rings look dirty. Rinse bio media in a bucket of water you just siphoned out of the tank, swish gently, put it straight back. Never under the tap, never in hot water, never scrubbed clean.
Replacing everything at once is the second. Change the mechanical stage as often as you like, but only ever replace part of the biological media at a time, and leave the new material in alongside the old for a few weeks so it colonises before the old comes out.
Cartridge dependency runs underneath both. Manufacturers sell monthly cartridges. Monthly cartridge changes remove a large share of your filter’s bacteria, which is why some tanks show a small ammonia blip a few days after every service. If your filter takes cartridges, cut the media basket into a sponge-and-ceramic setup instead and stop buying them.
Under-filtering is the actual risk in most tanks. Over-filtering is a flow problem you can solve with a valve and a sponge. Under-filtering is a stability problem you solve only by buying another filter, or by keeping fewer fish. Given the choice, buy the bigger one.
A quieter failure is the filter that runs perfectly and gets ignored. Flow drops gradually over eight weeks as foam loads with detritus, nobody notices because the change is slow, and by month three the tank is running at half the throughput its owner believes it has. Set a repeating reminder. A simple timer on the phone during a service keeps you from leaving media sitting in a bucket of cooling water for an hour, which is its own way to lose a colony.
Last, the power cut. A canister that sits unpowered for more than a few hours goes anaerobic, and restarting it can push a slug of foul water into the tank. If the power has been off overnight, open the canister, tip it out, rinse the media in tank water, and restart it clean rather than just flicking the switch.
Maintenance intervals worth keeping
| Task | Interval | Notes |
|---|---|---|
| Rinse pre-filter sponge | Weekly, with the water change | Tank water only, 20 seconds |
| Rinse mechanical foam | Every 2-4 weeks | Squeeze in old tank water until flow returns |
| Replace fine floss | Every 2-4 weeks | This one is genuinely disposable |
| Rinse biological media | Every 2-3 months, gently | Only if flow has dropped, never in tap water |
| Replace part of bio media | Only when it physically breaks down | Never more than a third at once |
| Clean impeller and shaft | Every 3-6 months | Slime buildup here is the usual cause of noise |
| Replace carbon | Every 3-4 weeks if used at all | Exhausts quickly, optional in most tanks |
| Check hoses and O-rings | Every 6 months | Silicone grease on canister seals prevents leaks |
Filter servicing and water changes belong on the same day but not the same hour. Doing both at once removes bacteria and dilutes the water column simultaneously, which is rarely a problem in a mature tank and occasionally is in a young one. Alternate them if your tank is under three months old. The full routine sits in our water change guide.
Marine and reef systems work to different assumptions again, with protein skimmers, sumps and live rock doing much of what a freshwater filter does, and dosing calculations that belong in the Reef Dosing Calculator rather than here.
Frequently asked questions
How many times per hour should an aquarium filter turn over the tank?
Four to six times the tank volume per hour is the standard target for a community tank, measured in real flow rather than the number on the box. Messy fish such as goldfish or larger cichlids do better at six to ten times. Low-flow species like bettas are happier at two to four times, and heavily planted tanks can sit lower still.
Is my filter too strong for my betta?
If your betta is being pushed around, hiding behind decor permanently, or resting on the substrate rather than swimming, yes. Long-finned Betta splendens tire quickly against current. Baffle the outflow with a piece of coarse sponge or a plastic bottle deflector, aim the flow at the glass instead of down the tank, and add floating plants to break up surface movement.
Canister or hang-on-back filter, which should I buy?
Hang-on-back filters are cheaper, easier to service and fine up to roughly 55 gallons (210 litres). Canisters hold far more media, run quieter, hide in the cabinet and suit larger or heavily stocked tanks, but they cost more and a monthly service takes 30 minutes rather than five. For most community tanks under 40 gallons a good hang-on-back is the sensible choice.
Is a sponge filter enough for a 20 gallon tank?
For a lightly stocked 20 gallon (75 litre) tank with small fish, yes, particularly a large dual sponge on a decent air pump. Sponge filters give excellent biological filtration and gentle flow, which suits shrimp, fry and bettas. What they do not do is trap fine particulate well, so the water may look slightly less polished than it would with a hang-on-back.
What order should aquarium filter media go in?
Water should meet mechanical media first, then chemical media if you use any, then biological media last. Coarse sponge or floss catches debris so it never reaches and clogs your ceramic rings or bio-foam. In a canister that usually means coarse foam at the bottom, fine foam above it, then bio media, with a polishing pad at the very end.
Can you over-filter an aquarium?
Biologically, no. Extra media capacity is only ever an advantage. The real limit is flow: too much current stresses low-flow species, blows substrate around and can strip carbon dioxide from a planted tank. If a filter is rated well above your tank, keep the media and dial the flow back with a valve, a spray bar, or a sponge over the outflow.
What actually changes for you
Work out your real water volume, pick a multiplier that matches your fish rather than your furniture, divide by 0.7 for a hang-on-back or 0.6 for a canister, and buy something at or above that rated number with the biggest media basket in its price band.
Then stop thinking about GPH. Once the filter is in, the numbers that determine whether it is doing its job are ammonia, nitrite and nitrate, tested with a liquid kit, not the flow figure on the packaging. A filter that keeps ammonia and nitrite at zero is correctly sized regardless of what its turnover ratio works out to, and a filter that does not is undersized regardless of how good the ratio looks on paper.