I am old enough to have run under-gravel filters when they were just what a filter was. A slotted plastic plate under the gravel, a couple of lift tubes, an air pump chugging away in the corner, and a tank of guppies that outlived the family cat. Then the hobby moved on. Canisters arrived, hang-on-backs got cheap, and somewhere along the way the under-gravel filter went from standard kit to the thing you are gently warned against by every shop assistant and half of YouTube. “Nitrate factory.” “Anaerobic dead zones.” “Antiquated.” I nodded along for years.

Then I read through David Bogert’s write-up on aquariumscience.org, which argues, with a fair bit of testing behind it, that almost everything the hobby now believes about under-gravel filters is wrong. I found myself agreeing more than I expected. So I am doing the honest thing: setting up a dedicated tank to run one properly, leaving it alone, and reporting back over months and years. More on that at the end. First, the case.

What an under-gravel filter actually is

An under-gravel filter, or UGF, is about as simple as filtration gets. A perforated plate sits on the tank bottom, the gravel goes on top, and water is pulled down through the gravel, under the plate, and back up through a lift tube. That is the whole machine. There is no cartridge, no impeller buried in a canister, nothing to replace.

The clever part is what the gravel becomes. Every grain is colonised by the same nitrifying bacteria that live in your filter media, turning fish waste ammonia into nitrite and then nitrate. A couple of inches of gravel across the whole footprint of the tank is an enormous surface area, far more than a sponge or a tray of ceramic rings. Bogert’s own comparison of filter types puts the UGF second only to a sump for sheer volume of biological media. You are, in effect, using the entire floor of the tank as your biofilter.

The brown gunk is the point, not the problem

Here is where I have to unlearn something, and where most keepers get it exactly backwards. Pull up a handful of gravel from an old UGF and it comes up brown, slimy and faintly whiffy. Every instinct says clean it. That instinct is the single biggest reason under-gravel filters get a bad name.

That brown material is not trapped waste. Bogert sent brown water from a 100-gallon tank through a fine filter sock after six months of running and recovered, by his account, only 2 to 5 teaspoons of a gravy-like material. He describes it as long-chain polysaccharides, the sugars that beneficial bacteria secrete as they build their colony. In other words it is mostly the living biofilm doing the filtering, not muck sitting there rotting. Swiss Tropicals, whom he quotes, put it plainly: the brown filter sludge is for the most part alive.

Clean it out and you throw away the bacteria you spent weeks growing during cycling. Do it hard and often enough and the filter genuinely stops working, at which point the critic points at your struggling tank and says “see, under-gravels don’t work.” They were right about the symptom and wrong about the cause. The filter did not fail. It was assaulted. This is the same mistake as the deep-clean disaster I wrote about in the maintenance schedule, just concentrated into the substrate.

The myths, taken one at a time

“It’s a nitrate factory.” Every effective biofilter is a nitrate factory. That is what nitrification does: ammonia in, nitrate out. A canister does exactly the same thing. It is basic conservation of matter, and the nitrate leaves the tank the same way regardless of filter, in a bucket at water change time. Singling out the UGF for this is nonsense.

“It’s antiquated.” As Bogert asks, since when is something bad simply because it is old? A design being from the 1970s tells you nothing about whether it works. Air itself is quite old and I still breathe it.

“Anaerobic dead zones form under the plate.” This is the scary one, and it is the one I was most worried about. Bogert argues that genuinely anaerobic, hydrogen-sulphide-producing zones need gravel of the order of two feet deep, a decade old, with no flow through it. At home-aquarium depths of one to three inches with water constantly drawn through porous gravel, he says it simply does not happen. My old tanks never gassed a fish, for what that is worth.

“They clog.” His claim is stark: never, in 54 years, if you leave them alone. The clogging people report is, in his telling, almost always self-inflicted by cleaning, or by using the wrong substrate.

Substrate: the one thing you must get right

This is where a UGF actually can fail, so it matters. The water has to flow down through the gravel, which means the gravel has to stay open.

  • Use natural aquarium gravel around 3 to 5 mm, roughly 3/16 to 1/4 inch. Two millimetres is about the fine limit. This is coarse enough for water to move through freely and hold its structure for years.
  • Depth of about 1 to 3 inches. An inch is the minimum for coverage; up to three or four inches is fine for heavier stocking and gives more media.
  • Avoid sand and anything under 2 mm. Fine sand packs down and seals the plate. It can be made to work over a weed-mat barrier, but Bogert is clear that this clogs far more readily, especially under heavy bioload or low-protein food.
  • Avoid blasting media and aquasoil on the plates. Sharp abrasive media and enriched aquasoils are both trouble. Aquasoil in particular breaks down into a clay-like sediment over time and will clog the plates, so keep it in a separate container or off the UGF entirely.

If you want to size a substrate order before committing, the substrate calculator will turn tank dimensions and depth into a bag count, and it is worth doing because good gravel is not cheap any more.

Flow: powerheads or air, your choice

Two ways to move the water. Air-driven lift tubes are the classic setup: an air pump, an airstone in the tube, bubbles rising and dragging water up with them. It is cheap, and small bubbles around a tenth of an inch across lift more water than big lazy ones. It also aerates beautifully, which is half the reason old UGF tanks were so stable.

Powerheads are the alternative and, honestly, the nicer option now. A small 4 to 5 watt powerhead on the lift tube is plenty, runs near silent, and Bogert reckons it is about 20 percent more efficient than an air lift. Not a huge margin, but quieter is quieter. Do not go mad on flow rate; gentle and constant beats a torrent.

Two things not to do. Do not plumb a canister filter in series with the plates. If the canister springs a leak, and canisters do leak, you can siphon the entire tank onto the floor and kill everything. And reverse-flow, where water is pushed up through the gravel instead of pulled down, is not worth the bother in his view; it reduces surface aeration for very little gain.

Maintenance, or the lack of it

This is the part that sold me, because I am fundamentally lazy about the right things. A properly run UGF is close to the lowest-maintenance biofilter there is. You still do your weekly water change and you still vacuum the open surface of the gravel where visible waste settles, exactly as in the normal routine. What you do not do is dig down and strip the bed. Bogert says he has not vacuumed deep into gravel in 54 years. If you ever feel the bed is genuinely tired, his method is to stir it into a brown soup and change most of the water once every couple of years, not to scrub it monthly.

The reader comments under his articles are worth a read on their own, and they are strikingly consistent. Long-time keepers in the comments report 35, 50, even 55 years on under-gravels with never a clog and crystal-clear water. One describes a planted 180-gallon tank running on nothing but a UGF and a single AquaClear powerhead. A Singapore hobbyist says the way old-timers recognise each other is the shared maxim to just leave the bloody thing alone; the shops there stopped stocking them not because they failed but because customers kept cleaning them to death. Another keeper mixed a UGF section with a sand cap in a tray and had clear water within three days. It is anecdote rather than data, but it is a lot of anecdote pointing the same way.

The honest against

I am not going to pretend this is settled gospel, because it is not, and the mainstream hobby disagrees for reasons that are not purely commercial. A UGF ties your biofilter to your substrate, so you cannot easily rescape without disturbing it. Big rooted plants and substrate-shifting cichlids can heave the plates up and expose them; Bogert’s answer is simply to cover them back over, and he says operation is unaffected, but it is a faff. Heavy stocking fed on low-protein food gives a poor carbon-to-nitrogen ratio that can overload any filter, UGF included. And you are committing to a specific substrate for the life of the tank. For a keeper who loves aquascaping and aquasoil, none of this fits. For someone who wants a stable, cheap, forget-about-it biofilter, it fits very well.

Bogert makes a further point I take seriously: the people telling you UGFs are obsolete are, quite often, the people selling canisters. There is a genuine commercial incentive to disparage a filter that costs twenty quid and lasts thirty years. That does not make him automatically right, but it is worth holding in mind when the received wisdom is unusually confident.

The experiment: a UGF tank I am starting now

So rather than argue about it, I am going to test it. I am setting up a dedicated tank on an old-fashioned under-gravel plate, and I am going to run it as a proper long-term experiment on this site.

The plan, following the specifics above: a plate across the full base, roughly 2 to 3 inches of natural 3 to 5 mm gravel, a small 4 to 5 watt powerhead on one lift tube. It will run on my usual hard, alkaline tap water, around 22 to 23 dGH, which suits the hard-water livebearers and Rift Lake stock I already keep, and it may well end up as another home for shell-dwelling cichlids. Stocking will be moderate, feeding sensible, and crucially I am going to do the hardest thing of all: nothing. Weekly water change, surface vacuum of the open gravel, and otherwise leave the bed strictly alone. No digging, no deep cleans, for years if it lets me.

What I want to know is the stuff you cannot get from a forum argument. Does it really stay clear? Does it clog? Do the fish do well over the long haul? How does the maintenance load actually compare to my sponge and canister tanks once you count the hours? I will log water parameters, the honest state of the gravel, and how much work it genuinely takes, and I will report back at intervals here. If it turns into a smelly clogged disaster in eight months, you will read about that too. That is rather the point of running the test instead of just having an opinion.

If you want to set one up alongside me, size the tank and its stock first with the volume and stocking calculators, and check the running cost of a small powerhead against your other kit with the running cost calculator. The whole appeal of a UGF is that those numbers come out low. Now to find out whether the low-maintenance promise survives contact with a real tank.