In-Ground Pool Losing Water: Find the Leak Before You Buy Anything
You're seeing: You're topping the pool up more than once a week, the water bill jumped, or the level is visibly below the skimmer mouth.
First rule out evaporation, because most 'leaks' are not. A quarter inch a day is most likely evaporation; half an inch a day is worth investigating. Confirm with the BUCKET TEST — a 5-gallon bucket filled to about an inch from the top, set on a step so it is immersed a few inches, water marked inside and out, pool at normal level and the pump running its usual schedule. If both drop the same, it is evaporation. AND RUN IT LONGER THAN 24 HOURS: owners consistently report the two levels looking identical at a day and clearly separated at three, so a one-day test can clear a real leak. Then use the two free diagnostics before buying anything. THE LEVEL IT STOPS AT IS THE HEIGHT OF THE LEAK — stops at the skimmer mouth, suspect the skimmer throat or faceplate; stops at a return, suspect that fitting; stops at the light, suspect the niche or its conduit; keeps going to the floor, suspect the main drain or the plumbing; never stops, suspect the shell. PUMP ON VERSUS OFF SPLITS PLUMBING FROM STRUCTURE — faster with the pump ON points at the pressure side (returns and the pipe run to them, now under pressure), faster or equal with it OFF points at suction side or the shell itself. Only then does the repair matter, and the product is chosen by CONSTRUCTION not by the hole: a VINYL liner takes a vinyl patch kit, most of which work underwater, while CONCRETE, GUNITE and FIBERGLASS take a two-part epoxy putty rated for underwater cure. Epoxy putty is not for vinyl and a vinyl patch will not hold on plaster. TWO THINGS THE LABELS DO NOT TELL YOU, both from reading the owner record rather than the box. On EPOXY PUTTY the repair is ruined by UNDER-MIXING, not by water: 'still pliable after 24 hours' is a two-part epoxy that never reacted, so knead until the mix gets WARM, and one owner reports that drying the area out for days made it worse while a little water made it behave. On VINYL, tape and a glue-and-patch kit are not equivalent under water — five separate owners report tape peeling off a submerged liner within a day, while five of seven owners who applied a glue-and-patch kit submerged report it sealing, and they all describe the same technique of two coats and a thirty-second press. One hard stop on fiberglass: if the laminate under the gelcoat is soft, delaminated or water-saturated, or water is actively streaming, that is not a putty job — stop and call someone.
Before anything else: most pools that are “leaking” are evaporating. Water loss is the single most over-diagnosed pool problem, and the fix for a misdiagnosis is expensive — people replace liners and hire detection services for a pool that was behaving normally in August.
So this guide runs in the order that costs least: rule out evaporation, then let the pool tell you where the leak is using two tests that cost nothing, and only then buy a repair product — chosen by construction, not by the size of the hole.
Step 1: is it a leak at all?
| Daily loss | Read it as |
|---|---|
| ~¼ in./day | Most likely evaporation |
| ~½ in./day | Worth investigating |
| More, or rising | Test properly, today |
Those are rules of thumb from Florida leak-detection work, and they are ranges because evaporation is driven by things that move week to week: air and water temperature, humidity, wind across the surface, sun, and whether the pool is covered. Anything that aerates increases it — a fountain, a spillover, deck jets. That is precisely what aeration does, and it is the same mechanism covered in water features and your pool.
The bucket test, done properly
- Fill a 5-gallon bucket to about an inch from the top, with pool water.
- Set it on a pool step so it sits immersed a few inches.
- Mark the level inside the bucket and the pool level outside it.
- Leave the pool at its normal level, pump on its usual schedule.
- Compare the two drops.
Equal drop = evaporation. Bigger drop outside the bucket = a leak, and the difference is roughly your leak rate.
Pool water and immersion both matter for the same reason: they keep the water inside the bucket at the same temperature as the water outside, and temperature is what drives evaporation. A bucket of cold tap water sitting on the deck measures nothing useful.
The refinement nearly every instruction omits: run it longer than 24 hours. Owners consistently report the two marks looking essentially identical after a day and clearly separated by day three. A one-day test can clear a slow leak that is genuinely there. An ambiguous result is not a pass — it is a reason to keep going.
Step 2: the two free diagnostics
Do these before buying anything. Between them they usually name the suspect.
The level it stops at is the height of the leak
Stop topping the pool up. Let it drop on its own and note where it stabilises.
| It stops at | Suspect |
|---|---|
| The skimmer mouth | Skimmer throat, faceplate gasket, or the joint where a plastic skimmer meets the shell |
| A return fitting | That fitting |
| The light | The niche — or the conduit out of the back of it, an easily missed path where water leaves through the conduit rather than the wall |
| The floor | Main drain, or plumbing below |
| It never stops | The shell, or a line under the floor |
This costs nothing but patience and it eliminates most of the pool.
Pump ON vs pump OFF splits plumbing from structure
Compare the loss rate running against not running.
- Faster with the pump ON → the pressure side. Return plumbing is only pressurised when the pump runs, so a pressure-side leak shows up on the pump’s schedule.
- Faster or equal with the pump OFF → the suction side, or the shell itself.
Then dye, and only where the first two tests pointed
With the pump off and the water still — circulation scatters dye and wastes your afternoon — release a small amount of leak-finding dye or food colouring right beside the suspect surface. At a hole the dye draws in as a thin steady thread instead of dispersing.
Search the band the level test identified, all the way around, before looking anywhere else. Random dye testing is how people spend a weekend learning nothing.
Look outside the pool too. A persistent wet patch or a strip of unusually lush grass on one side is evidence. So is a soft or washed-out spot under an above-ground wall. And check the equipment pad first — pad drips are far easier to fix than anything in the shell, and they get blamed on the pool constantly.
This is not the same as a pump sucking air
A suction-side air leak is air getting IN; a water leak is water getting OUT. They get confused constantly and they are diagnosed in opposite ways.
| Air leak | Water leak | |
|---|---|---|
| Symptom | Bubbles/froth in the pump basket, air at returns, pump won’t prime | Level drops |
| Water loss | Often none | That is the symptom |
| Found by | Pouring water over fittings while the pump runs | Level test + dye, pump off |
If your symptom is bubbles, you want pump and filter troubleshooting instead. If you have both, fix the air leak first — it is usually cheaper and more accessible, and it occasionally turns out to have been the whole story.
Above-ground pools have their own guide
If your pool is above-ground, stop here and read above-ground pool losing water instead. The diagnosis overlaps — evaporation, the bucket test, the level test, dye — but two things differ enough to need their own page. A leak on an above-ground pool washes out the base the wall stands on, which turns a water bill into a structural clock, and the repair question is entirely about vinyl, where the tested answer is not the one most guides give.
In-ground vinyl-liner pools stay here for diagnosis and take the vinyl repair in the table below.
In-ground concrete and gunite
Finding is the same: level test, pump on/off, then dye. What differs is where to look and what to use.
Look at the skimmer throat (the most common structural leak on a gunite pool), the light niche and its conduit, return fittings, tile-line cracks, and any crack that holds dye.
The repair: a two-part epoxy putty rated to cure underwater, kneaded and pressed into the crack or around the fitting.
The mistake that ruins this repair is mixing, not water. We read the owner record on both sides. The unhappy reviews say the putty dissolved, flaked, or was “still pliable after 24 hours” — and that is the signature of a two-part epoxy that never reacted, not of a wet surface. The recent five-star reviews are full of skimmer-throat repairs made without dropping the pool an inch, and one owner reports that drying the area out for days made it worse: it only came together when water reached it.
So do what the owners who succeed do:
- Knead the two parts until the mix gets warm. One owner sets a five-minute timer and does not stop when it merely looks one colour — the warmth is the reaction, and it is the only reliable tell.
- Clean the surface first. A magic eraser or a Scotch-Brite pad, not a rinse.
- Leave the pump off while it sets.
The stop rule: a hairline crack that holds dye is a putty job. A crack that moves, runs the height of a wall, or reopens after repair is telling you the shell is moving. That wants an engineer, not a tube of anything.
In-ground fiberglass
Fiberglass leaks are usually at fittings and penetrations rather than in the shell, so start there.
Where the shell itself is damaged, the question is depth:
- Gelcoat only — a chip, surface crack or scratch that holds dye, with sound laminate underneath → a wet-surface-rated epoxy or gelcoat repair kit. The two-part putty above is rated for fiberglass as well as concrete and gunite, and owners use it on fiberglass steps and skimmer housings. Colour matching is the hard part; adhesion is not.
- Laminate soft, delaminated or water-saturated, or water actively streaming → stop. This needs dewatering and proper lamination repair. Putty pressed over saturated laminate traps water inside the shell and makes the eventual repair worse and dearer.
The buying rule: the product is chosen by construction
This is where money gets wasted, because the two main products look interchangeable on a shelf and are not.
| Your pool | Use | Do not use |
|---|---|---|
| Vinyl liner (all above-ground, vinyl in-ground) | Vinyl patch kit — vinyl + underwater-rated adhesive | Epoxy putty. It is not made for vinyl |
| Concrete / gunite | Two-part epoxy putty, underwater cure | A vinyl patch. It will not hold on plaster |
| Fiberglass | Wet-surface epoxy or gelcoat kit — gelcoat-depth damage only | Anything, if the laminate is saturated |
| Any construction, at a fitting | Pool-rated underwater adhesive or putty around the fitting | Shell products applied to the shell |
What to check on the label before you buy: that it is rated for underwater application and cure if you are not draining, and that the construction it names matches yours. A tube that says “concrete, gunite and fiberglass” is telling you plainly that it is not for your liner.
Our picks map straight onto that table: the two-part epoxy putty for concrete, gunite, plaster and fiberglass; the glue-and-patch kit below for vinyl; and a pour-in sealer as the last rung if the leak stays hidden. Each was chosen on a paced read of its owner record, not on a roundup.
If your pool is vinyl-lined
Glue and patch, not tape. We read the owner record across the three repair types and this is the only one whose underwater claim holds up: five of the seven owners who wrote specifically about applying it submerged say it sealed, on the cleanest base in the category. The successes all describe the same method — two coats of glue, press hard for a full thirty seconds, round the patch so it has no corners to peel from, then leave the water still.
If the liner is dry — a drained pool, or a tear above the waterline — a solvent vinyl cement is the stronger repair. The Boxer kit below welds rather than sticks and its owners who patched dry report it holding a season; its critical tail is entirely the underwater case (patches that roll up, glue that floats off, one liner eaten through), which is why it is not the pick above. Choose by whether you can reach the hole dry, not by price.
Tape has a place, and it is above the waterline — the full argument, and the owner evidence behind it, is in above-ground pool losing water, which applies to any vinyl liner.
If you have run every test and still cannot find it
Poured in and circulated, it seals leaks up to about an eighth of an inch, including plumbing leaks that the level test can point at but no patch can reach. It is the one product here that ignores the construction rule — concrete, fiberglass, vinyl and PVC all take the same bottle.
Buy it knowing what it is. Its critical share is 12%, three times the epoxy putty’s, and the complaint is almost always the same one: the leak came back. Even a five-star owner describes re-dosing an unfixable spa leak every four to six months. That makes it a way to reach a planned repair instead of an emergency one — not a repair.
When to stop
Call a leak-detection service when the level test says the leak is below the floor or in the plumbing, when a crack moves or reopens, when fiberglass laminate is saturated, or when you have run the two free tests and the pool still will not tell you anything. Pressure-testing individual lines needs equipment and access most owners do not have, and a plumbing leak under a deck is not a DIY afternoon.
The tests above are worth running first regardless — arriving with “it stabilises two inches below the skimmer and it is faster with the pump running” turns an open-ended hunt into a short visit.
Frequently asked
How much water loss is normal for a pool?
Roughly a quarter of an inch a day is most likely evaporation rather than a leak, and about half an inch a day is worth investigating. Those are rules of thumb from Florida leak-detection work, and the reason they are ranges rather than numbers is that evaporation is driven by things that change week to week: air temperature, water temperature, humidity, wind across the surface, direct sun, and whether the pool is covered. A heated pool in a dry breeze can lose far more than an unheated one in still humid air, and any water feature that aerates — a fountain, a spillover, deck jets — increases loss further because that is exactly what aeration does. So do not diagnose from the number alone. The number tells you whether to bother testing; the bucket test tells you whether there is actually a leak. And if you have just added a water feature or started running the heater, expect the baseline to move on its own.
How do I do the bucket test properly?
Fill a 5-gallon bucket to about an inch from the top with pool water, then set it on a pool step so it sits immersed a few inches — using pool water and immersing the bucket both matter, because they keep the water inside at the same temperature as the water outside, and temperature is what drives evaporation. Mark the water level inside the bucket and the pool level outside it, using tape or a grease pencil. Leave the pool at its normal level and run the pump on its usual schedule, because changing either changes what you are measuring. After the test, compare the two drops. Equal means evaporation. A bigger drop outside the bucket than inside means a leak, and the difference is roughly your leak rate. The one refinement most instructions omit: run it longer than 24 hours. Owners repeatedly report the two marks looking essentially identical after a day and clearly separated by day three, so a single-day test can clear a slow leak that is genuinely there. If the result is ambiguous, that is not a pass — it is a reason to keep going.
Where is my pool leaking?
Let the pool tell you, because the level it settles at is the height of the leak. Stop topping it up and let it drop on its own. If it stabilises at the skimmer mouth, suspect the skimmer throat, the faceplate gasket or the joint where a plastic skimmer meets the shell. If it stops level with a return, suspect that fitting. If it stops at the light, suspect the niche or the conduit running out of the back of it — a common and easily missed path, because water can leave through the conduit rather than the pool wall. If it runs all the way down to the floor, suspect the main drain or the plumbing below. If it never stabilises at all, you are looking at the shell or a line below the floor. The second free test separates plumbing from structure: compare the loss rate with the pump running against the pump off. Faster with the pump ON points to the pressure side, because the return plumbing is only pressurised when the pump runs. Faster or equal with it OFF points to the suction side or to the shell itself. Between those two tests you can usually name the suspect before spending anything.
My pool is above-ground — is this the right guide?
No, and the reason is not just tidiness. On an above-ground pool the escaping water goes into the prepared base the wall stands on, washes it out, and takes the pool's footing with it — so a slow leak there is the start of a structural failure rather than only a water bill, and it wants acting on faster than the same leak in an in-ground pool. The repair question is different too: an above-ground pool is entirely vinyl, and the tested best answer for patching vinyl is not the one most guides give. The diagnosis does overlap — evaporation first, the bucket test run for three days rather than one, then letting the level tell you the height before you dye anything — so nothing you read here is wasted. But go to [above-ground pool losing water](/pool-school/above-ground-pool-leak/) for the search order, the base-washout warning signs, and what actually patches a liner. In-ground VINYL-liner pools are covered here, since they share the in-ground diagnosis and the same vinyl repair.
What do I use to fix the leak once I find it?
The product is chosen by CONSTRUCTION, not by the size of the hole, and the two main options are not interchangeable. A VINYL liner — every above-ground pool and any in-ground vinyl pool — takes a vinyl patch kit: a piece of liner vinyl plus an adhesive rated for underwater use, and most modern kits can be applied submerged so the pool does not need draining. Round the corners of the patch before applying it, because square corners peel first. CONCRETE, GUNITE and FIBERGLASS take a two-part epoxy putty rated to cure underwater, which you knead and press into the crack or around the fitting. That putty is explicitly not for vinyl liners, and a vinyl patch will not hold on plaster, so buying the wrong one wastes both the money and the trip. For fittings and returns on any construction, a pool-rated underwater adhesive or putty around the fitting is usually the fix rather than anything applied to the shell. Two things the labels leave out, both from the owner record rather than the box. On the putty, the repair is ruined by under-mixing rather than by water — knead the two parts until the mix gets warm, because that warmth is the reaction and it is the only reliable sign it will cure; owners who report it dissolving or staying soft are describing a mix that never got there. On vinyl, tape and a glue-and-patch kit are not interchangeable below the waterline: owners report tape letting go of a submerged liner within a day, while the glue-and-patch kit holds for most who use it, provided they apply two coats and press hard for a full thirty seconds. And be honest about what a patch is: on a liner past its life or a shell with structural cracking, a patch is a way to buy a season, not a repair.
Can a fiberglass pool leak be patched, or does it need a professional?
It depends entirely on how deep the damage goes, and there is a clear stop rule worth knowing before you start. If the damage is confined to the gelcoat — a chip, a surface crack, a scratch that holds dye — and the fiberglass laminate underneath is intact and sound, a wet-surface-rated epoxy or a gelcoat repair kit will handle it, and colour-matching is the hard part rather than the adhesion. But if the laminate under the gelcoat is soft, delaminated or water-saturated, or if water is actively streaming rather than seeping, stop. That is not a patch job. It needs dewatering and proper fiberglass lamination repair, and putty pressed over saturated laminate traps water inside the shell and makes the eventual repair worse and more expensive. The same logic applies to concrete: a hairline crack that holds dye is a putty job, while a structural crack that moves, runs the height of the wall, or reopens after repair is a sign the shell is moving and wants an engineer rather than a tube of anything.
Is my pump sucking air the same problem as a leak?
No, and they are opposite problems that get confused constantly. A suction-side AIR leak is air getting IN to the plumbing before the pump — you see bubbles or froth in the pump basket, air at the returns, or a pump that will not hold prime, and the pool may lose no water at all. A water leak is water getting OUT. They can share a location, since a bad union or a failed pump-lid o-ring can do both, but they present differently and you diagnose them differently: an air leak is found by pouring water over fittings while the pump runs and watching for the froth to clear, whereas a water leak is found by the level test and dye with the pump off. If your symptom is bubbles rather than a dropping level, you are in the wrong guide — [pump and filter troubleshooting](/pool-school/pump-and-filter-troubleshooting/) is the one you want. If you have both, fix the air leak first, because it is usually the cheaper and more accessible of the two and it sometimes turns out to have been the whole story.