What size filter do I need for my tank?
One rated for 4–6 tank volumes an hour for an ordinary community tank, 8–10 for a planted tank that needs the CO₂ moved around, 5–10 through the sump for salt water. A standard 55 gallon community tank holds about 205 liters as kept, so it wants a filter rated 220–320 GPH — the middle of that for moderate stocking, the top for goldfish and cichlids. Those multiples are quoted against the number printed on the box, because that is the only number anyone had when the rule was written, and it is the basis this page sizes on. What the box number actually moves once the baskets are full is the next section, and it is why comparing a canister with a hang-on-back by rating alone goes wrong.
| Tank | Community (4–6×) | Planted (8–10×) | Sump, salt (5–10×) | |
|---|---|---|---|---|
| 5.5 gallon | 20–30 GPH | 40–50 GPH | 25–50 GPH | |
| 10 gallon | 40–55 GPH | 75–95 GPH | 50–95 GPH | |
| 15 gallon | 55–80 GPH | 110–135 GPH | 70–135 GPH | |
| 20 gallon long | 70–105 GPH | 135–170 GPH | 85–170 GPH | |
| 20 gallon high | 75–110 GPH | 150–185 GPH | 95–185 GPH | |
| 29 gallon | 105–160 GPH | 210–260 GPH | 130–260 GPH | |
| 37 gallon | 130–195 GPH | 260–330 GPH | 165–330 GPH | |
| 40 gallon breeder | 170–250 GPH | 340–420 GPH | 210–420 GPH | |
| 55 gallon | 220–320 GPH | 430–540 GPH | 270–540 GPH | |
| 65 gallon | 260–390 GPH | 520–650 GPH | 320–650 GPH | |
| 75 gallon | 300–450 GPH | 600–750 GPH | 370–750 GPH | two filters |
| 90 gallon | 340–520 GPH | 690–860 GPH | 430–860 GPH | two filters |
| 120 gallon | 460–690 GPH | 920–1,150 GPH | 570–1,150 GPH | two filters |
| 125 gallon | 450–670 GPH | 900–1,100 GPH | 560–1,100 GPH | two filters |
| 150 gallon | 610–910 GPH | 1,200–1,500 GPH | 760–1,500 GPH | two filters |
| 180 gallon | 720–1,100 GPH | 1,450–1,800 GPH | 900–1,800 GPH | two filters |
| 210 gallon | 840–1,250 GPH | 1,700–2,100 GPH | 1,050–2,100 GPH | two filters |
Rated (box) flow, water as kept — an inch below the rim over gravel; switch the tool to liters for L/h. Salt water's 20–40× is powerhead circulation, sized separately. "Two filters" from 75 gallons.
Manufacturers rate flow at the pump, no media, no hose, no spray bar. Hobbyists who have measured an Eheim 2217 — rated 264 GPH — get 140–165 with media and a spray bar, 53–63 % of the rating; an Aqua One rated 1,400 L/h delivers about 900. A hang-on-back does a little better (a 200 GPH unit moves 100–140) because the water has less far to go. So the 55 gallon tank's "300 GPH" canister really turns it over about 3.1× an hour, not 5.5 — and that is the flow every working tank on the forums has always had, which is why the rule is left on the box basis rather than doubled. The tool prints both numbers so a canister and a hang-on-back can be compared on the same footing: a 300 GPH HOB moves more water than a 300 GPH canister, and holds about a fifth of the media.
Does a canister lose flow sitting under the tank? No — it's a closed loop
The head-height tables the calculators copy from pump catalogs do not apply to a canister filter. Its intake tube is a siphon: the water column standing in it pushes on the pump inlet with exactly the pressure the outlet has to overcome to lift the water back up. Net lift, zero — a canister on the floor under a five-foot stand sees the same head as one sitting beside the tank. What costs a canister flow is friction: every foot of hose, every elbow, the spray bar, and above all the media as it loads with muck. That is why a long hose run to a cabinet across the room matters and stand height does not, and why the flow you get in month three is the cleaning schedule, not the pump. Static head is real for a sump, where the return pump lifts to a free surface and every inch of rise takes flow off the curve — that is the return pump calculator's problem, and it is a different one.
How much filter media do I need?
More than flow. aquariumscience.org ran the test: doubling the flow through a canister added about 9 % to its capacity, because the bacteria live on the media surface and the water was already passing it more often than they needed. The spec that matters is the basket, and makers rarely print it — but three that are known line up: an Eheim 2217 holds 6 L on a 1,000 L/h rating, a SunSun HW-304B 9 L on 2,000, a Fluval FX6 about 16 L usable on 3,500 — close to 5 mL of media per L/h of rating, which is how the tool estimates a basket you can't find a figure for. The "300 GPH" canister above holds about 5.7 L, 2.8 % of the tank; a 300 GPH hang-on-back about 1.7 L. Fill the basket with something with surface — 30 PPI foam, K1, pot scrubbers, sintered glass — rather than ceramic rings, and buy the canister with the bigger baskets before the one with the bigger number.
| Filter | Really moves | Media per 1,000 L/h rated | Stand height | |
|---|---|---|---|---|
| Canister | 55 % of the box | 5 L | free | the most media per dollar; closed loop, so stand height is free |
| Hang-on-back | 65 % of the box | 1.5 L | free | short lift, little media; a cartridge is mostly floss |
| Internal power filter | 80 % of the box | 1 L | free | no hoses, no lift — but the least media |
| Sponge (air or powerhead) | 100 % of the box | — | free | flow is whatever drives it; the sponge is the media |
| Sump with return pump | 70 % of the box | — | costs flow | the return pump lifts to a free surface, so head really does apply |
Real-flow fractions from hobby measurements of named units (Eheim 2217, Aqua One Aquis 1250, 200 GPH hang-on-backs); media from the three published baskets. Sponges take whatever flow drives them. Coarse, and stated as such — a plug-in flow meter on the outlet beats every row.
The "20–40× for a reef" figure that turns up in filter calculators is circulation — powerheads pushing water around the rock — not water through a filter. Through the sump the consensus is 5–10×: enough for the skimmer and the media to see the water, slow enough that the overflow stays quiet and the return pump isn't an electricity bill. On a standard 120 gallon reef (434 L as kept) that is a return of 570–1,150 GPH and powerheads doing 2,300–4,600 GPH on top. The tool separates the two; the powerhead calculator sizes the circulation.
Canister, hang-on-back or sponge — which for my tank?
By media and by what it will let you do. A sponge is nearly all media and no flow — the best filter per dollar for a shrimp tank, a breeder, a quarantine or anything under 20 gallons, and useless at clearing water. A hang-on-back moves water well and holds almost nothing; it is a mechanical filter with a token biological stage, fine to 40 gallons if you stuff the box with foam instead of the cartridge. A canister is the only common filter whose media volume scales with the tank, which is why it is the default from 40 gallons up and the only sensible choice for a stocked 75. Internal power filters sit between sponge and hang-on-back. Whatever you buy, the flow number decides the turnover and the basket decides the capacity, and the tool shows you both for the one you are looking at.
Should I run two filters?
From about 75 gallons, yes — two smaller units rated to the same total. Three reasons, none about flow: a pump fails or an impeller jams and the tank still has a working colony; you can clean one this month and the other next, so the biofilter never takes the hit of a full clean; and a four- or six-foot tank gets an outlet at each end instead of a dead corner where cyanobacteria starts. The cost is two sets of hoses and two intakes in the glass. Below that size a single filter with a sponge prefilter on the intake does most of the same job, and the sponge is the spare colony.
My filter is the right size and the water is still dirty — why?
Because size is not the variable. Cloudy or particulate water is a mechanical problem — fine floss or 30 PPI foam in the last stage, cleaned weekly, clears it; a bigger pump pushing water past coarse sponge faster does not. Ammonia or nitrite with a "right-sized" filter is a colony problem — a new tank, an over-cleaned basket, or a cartridge thrown away with the bacteria on it. Dead spots are an outlet problem: point the output along the long axis, or add a powerhead. And a filter that was fine in January and isn't in June has simply loaded up: the real flow through a dirty canister is a fraction of the clean figure, and the cure is a rinse in tank water, not a bigger filter. The tool's "really moves" number is the clean one; take a third off it for the month before a clean.