My fish are dying but the water looks clean — why?

Ammonia and nitrite are invisible and odorless at killing doses. Water can look crystal clear while reading 2 ppm ammonia or 5 ppm nitrite. Only a liquid test kit sees new tank syndrome; clear water is not a test result.

Should I do a fish-in cycle or a fishless cycle?

Fishless, whenever it can be arranged: you grow the bacteria on pure ammonia or fish food with zero patients on the ward. Fish-in cycling forces hardy fish to endure the toxic peak so the tank can mature around them; it can work with daily testing and changes, but it is risk management, not best practice.

Fish are dying right now in my new tank — what do I do?

Water change first: 40-50% immediately with dechlorinated, temperature-matched water, then test ammonia and nitrite. Every daily 40% change roughly halves the reading. Stop feeding, add an airstone, and repeat until ammonia and nitrite both hold at 0 ppm.

Fish Keep Dying in a New Tank? New Tank Syndrome, Start to Finish

Q: Why do my fish keep dying in a new tank?

New tank syndrome: the beneficial bacteria that convert toxic ammonia to nitrite and then to far-less-toxic nitrate have not colonized the filter yet. Fish waste raises ammonia, a second bacteria population lags behind it raising nitrite, and both burn fish long before the tank 'looks' established. Deaths clustered in weeks 1-6 of a new aquarium are this pattern until a liquid test proves otherwise.

Q: How long does it take to cycle a new tank?

Roughly 4-6 weeks fishless with an ammonia source, sometimes 6-8 in cool water. Cycled means: ammonia 0 and nitrite 0 within 24 hours of dosing ammonia, with nitrate rising. Adding fish before that curve completes is the single biggest cause of new-tank deaths.

Q: Can new tank syndrome kill fish?

Yes — and it is the number one killer of new-tank livestock. The ammonia and nitrite spikes of an uncycled tank burn gills and blood chemistry at doses that are invisible and odorless, and fish that looked fine at pet-store pickup die within weeks 1-6. Every death in a tank under six weeks old is new tank syndrome until a liquid test proves otherwise, and the same spike can silently re-run itself in older tanks after a dead fish is missed, a filter is rinsed in tap water, or feeding is overdone.

Q: How long until a new tank is safe for fish?

4-6 weeks with a fishless cycle, sometimes 6-8 in cool water or with a big filter load. The calendar is not the finish line — the test is: dose the tank to 2 ppm ammonia and stock only when it converts to 0 ammonia and 0 nitrite within 24 hours while nitrate rises. Seeding media from an established tank can cut that to 2-3 weeks; adding fish on day 3 because the water 'looks clear' is how weeks 2-6 become a casualty list.

Last updated: November 2026

Act now — fish are dying in the water, not from it being "new"

Do a 40–50% water change right now with dechlorinated water matched to tank temperature within 1–2 °F. This is the rescue, before any diagnosis.
Test ammonia and nitrite with a liquid kit. Both must read 0 ppm. Any reading above zero in a tank under about six weeks old is your answer.
Stop feeding, add an airstone, and repeat the 40% change daily until ammonia and nitrite hold at zero. Every daily change roughly halves the reading.

Severe cases: consult an aquatic-experienced vet where available.

A new aquarium that keeps killing fish is almost never cursed, defective, or haunted — it is uncycled. The pattern is so reliable it has a name: new tank syndrome. Fish go in happy, the water looks sparkling, and then in the second to sixth week they gasp, clamp, go lethargic, and die one by one, usually the most sensitive fish first. This page explains the actual mechanism (the nitrogen cycle), shows you the day-by-day ammonia/nitrite/nitrate pattern so you can recognize exactly where your tank is on the curve, gives the honest fish-in vs fishless verdict, and ends with the water-change-first protocol that saves the fish you still have. If a fish is down to passive, drifting-at-the-surface territory, start with the first-hour rescue and come back.

What's actually happening: the nitrogen cycle in one paragraph

Every fish produces ammonia (through its gills and waste), and uneaten food decays into more. Two different wild bacteria populations clean this up in sequence, and neither one comes pre-installed in fresh tap water: population one (Nitrosomonas-type) eats ammonia and excretes nitrite — which is as toxic as ammonia to fish blood — and population two (Nitrospira-type) eats nitrite and excretes nitrate, which is mild at reasonable levels. Each population can only grow while its food is present, and each takes weeks to build to the size your fish load demands. That lag is the whole story: ammonia spikes before population one is big enough, nitrite spikes while population two hasn't even started, and fish live inside the toxic middle of a graph they can't see. Bacteria live mostly in filter media and substrate, which is why new filters, tap-water media rinses, and "deep cleans" reset the clock. The visual companion to this crash — milky white haze — is covered on the cloudy water decoder; the bacteria in that bloom are the reason the tank looks sick right as the invisible killers peak.

The day 0–42 timeline: what your tests should show

Days (typical fishless cycle)	Ammonia	Nitrite	Nitrate	What it means
Day 0–3	0 → rising	0	0	Setup or ammonia dosing begins. Nothing to see; bacteria populations are near zero.
Day 4–10	High (2–4 ppm)	0 → rising	0	Ammonia eaters are starting to colonize. A fish added today is swimming in the worst window. Milky water is common here.
Day 10–21	Peaking then falling	High (2–8+ ppm)	Still ~0	The nitrite gap — the classic kill zone. Nitrite is at its worst while nitrate hasn't started. This is where new-tank fish die of brown, chocolate gills.
Day 21–35	Falls to 0	Follows it down to 0	Climbing	Nitrate makers are catching up. The conversion is visibly accelerating: test water now clears an ammonia dose faster each week.
Day 35–42+	0 within 24 h of dosing	0 within 24 h of dosing	Rising steadily	Cycled. Ammonia 0 + nitrite 0 within a day, with nitrate rising, is the finish line — now add a small starter load, not a shopping-spree tank.

In a fish-in tank the same curve runs on a smaller scale — the fish themselves are the ammonia source — and it's grueling to watch: fish show clamped fins and lethargy around days 7–21 (see lethargic and not eating), gasping arrives around the nitrite peak (the gasping protocol), and deaths of the sensitive fish — neon tetras, wild discus, fancy guppies — come next. A tank can also decycle months later: a dead fish left in decor, a filter cleaned in tap water, or overfeeding can re-run the ammonia leg of this curve in an "old" tank, which is why veterans test after every loss.

9 reasons your new tank keeps killing fish

New-tank losses get blamed on bad luck, bad fish, or a cursed aquarium. They almost never are. Here is the full suspect list, ranked by how often it actually shows up at the autopsy — the first two are new tank syndrome itself, and numbers 3–9 are the partners in crime that make an honest beginner's tank lethal anyway.

An uncycled tank — the ammonia wave. The #1 killer, full stop: fish went in before the ammonia-eating bacteria exist, so their own waste accumulates as poison. The tells are dead-on-timing — losses starting in week 2, several fish off at once, clamped fins and lethargy first (what poisoning looks like). Fix: the water-change protocol below, every day, until ammonia reads 0.
Nitrite — the second wave. Just as ammonia falls, the nitrite spike crests days 10–21 and kills with the same invisible efficiency, browning the gills and locking oxygen out of the blood; gasping over clean-looking water is its signature (the gasping protocol). Fix: daily changes to keep it under ~0.5 ppm, plus 1 ppt aquarium salt to protect gills — no shrimp or scaleless fish, obviously.
Chlorine or chloramine in the water you're adding. Plain tap water burns fish gills outright — and quietly kills the very filter bacteria you're trying to grow, so a tank that "has been cycling for weeks" never actually moves. Chloramine is worse: it keeps releasing ammonia into your finished water for days. Fix: a proper conditioner (one that neutralizes chloramine) on every drop of tap water that enters the tank, and a liquid kit to prove it (kit picks).
Temperature shock and sloppy acclimation. Dropping a bagged fish into water more than a few degrees from its transport temperature — or floating the bag but never equalizing the water — produces acute shock: gasping, darting, then listlessness within hours, and chilled fish go on to catch ich. Fix: float to match temperature, then add cupfuls of tank water over 15–20 minutes before netting in; match change-water within 1–2 °F at every water change.
Overstocking on day one. The whole wish-list stocked at once means the bacteria that survived cycling face a bioload three times anything they've grown to handle, and the tank decycles under the load. It's the most common way a cycled tank starts killing anyway. Fix: stock a fraction of the final load, then add in steps every 2–3 weeks, checking totals against the stocking calculator.
Overfeeding. In a new tank every uneaten flake is pre-paid ammonia, and anxious beginners feed a lot; the food decays, feeds the bacterial haze on the glass (the milky-water decoder), and outpaces a filter colony that's still weeks from full size. Fix: feed once daily, only what vanishes in 60 seconds, and skip days entirely while toxins read above zero — established fish are never the emergency; the food they leave behind is.
Soap and residue contamination. New tanks are "cleaned" with detergent, new decor scrubbed with soap, hands washed with soap before reaching in, or aerosols sprayed over open water — all of which strip fish gills and crash the bacterial colony in one dose. Fix: new tanks and decor get hot water and a dedicated brush only; no chemical ever touches anything that gets wet with fish water.
A dead fish or hidden snail clump nobody removed. One mouse dead behind the filter housing, or a snail colony that died in a log, runs a continuous ammonia drip that looks exactly like a decycle — spikes that keep returning after every water change. Fix: find and remove the source (smell near the filter, inspect every hollow decor piece), then daily changes until readings hold at 0 — this same rule is why the emergency protocol says pull any fish that dies, immediately.
Sick fish from the store. Livestock that left the shop carrying ich, columnaris, or internal parasites often crashes in your tank within 1–2 weeks — the transport stress plus the new-tank spike does what the store's water was holding off. This is why deaths continue even in a perfectly cycled new tank. Fix: buy from high-turnover shops, inspect before paying (clamped fins, spots, sores: the spots guide), and observation-bag new arrivals away from the main tank for 2–3 weeks; a balance-losing, tilting arrival is covered on the swim bladder page.

Work the list top to bottom with a liquid test in hand: causes 1–2 are visible in one set of readings, cause 3 is visible in your routine, and 4–9 are visible in your habits. If more than one applies, it explains the losses; you never need a tenth theory.

Fish-in cycle vs fishless cycle — the verdict

Fishless wins; fish-in is damage control, not a method

Fishless cycling (recommended): fill, filter, heat, and dose ammonia (pure ammonium chloride, or a pinch of fish food daily) with no animals present. Test every 2–3 days; when 2 ppm ammonia converts to 0 ammonia / 0 nitrite within 24 hours with nitrate rising, do a big change and stock gradually. Cost: 4–6 weeks of patience. Benefit: zero patients on the ward — and every beginner-friendly species on the species pages arrives into safe water.

Fish-in cycling (only if fish are already there): it has worked for decades because hardy fish tolerate moderate toxin exposure better than marketing admits — but "tolerate" is doing heavy lifting. If you're mid-crash right now: test ammonia and nitrite every day, keep both combined under ~0.5 ppm with changes (up to 50% daily — matched temperature, dechlorinated, always), feed once every other day, and hold stocking at well under the calculator's ceiling. Hardy Danios and similar work here; a school of neon tetras does not. Never "cycle with a goldfish" or "the plants need fish" — both are folklore with a body count.

Emergency protocol: fish dying right now in a new tank

40–50% water change immediately, temperature-matched and dechlorinated, poured gently onto a saucer or plant. Halving ammonia and nitrite now is worth more than any bottle at the pet shop.
Test and write the numbers down: ammonia, nitrite, nitrate, temperature. The numbers tell you which leg of the curve is killing fish, and the trend over days tells you if the filter is establishing at all. No kit yet — buy a liquid kit; strips are how beginners mis-read this emergency (kit picks).
Nitrite above zero and fish gasping? Add 1 ppt aquarium salt (~3 g/L) to block nitrite uptake — only if no shrimp, snails, or scaleless catfish are in the tank. Full salt and aeration math on the gasping page.
Feed nothing while toxins read above zero, and remove any dead fish immediately — one rotting fish is an ammonia grenade for everyone left.
Repeat 40% changes daily until two consecutive days of 0 ammonia / 0 nitrite. Add nitrifying bacteria supplements only after changes, as an accelerant, never a substitute.
Then fix the root cause so it doesn't recur: fewer and smaller stocking steps, feed smaller amounts, weekly 25–40% changes, media rinsed only in old tank water. The day-by-day version of all this is the free new tank cycle checklist (PDF).

Prevention: the rules that end this question forever

Fishless cycle to 0/0 before any fish; test the finish line (conversion within 24 h), don't guess from the calendar.
Stock in steps: a fraction of final load at cycled day one, new additions every 2–3 weeks so bacteria grow ahead of the bioload.
Never rinse filter media in tap water; squeeze it in removed tank water at changes.
Test weekly for the first three months and after every behavior change; the targets are ammonia 0, nitrite 0, nitrate under 40 ppm.
Seed a new filter or tank with media from an established tank or a store's display sponge — the one legitimate shortcut to weeks of bacteria growth.
Check the Symptom Finder early: clamped fins and lethargy in a young tank is the crash announcing itself before the deaths start.

FAQs

Why do my fish keep dying in a new tank?

New tank syndrome: the bacteria that convert ammonia → nitrite → nitrate need weeks to colonize, and fish added early live inside the ammonia and nitrite spike. Deaths clustered in weeks 1–6 of a young tank are this pattern until a liquid test says otherwise.

The water looks crystal clear — how is it killing them?

Ammonia and nitrite are invisible and odorless at killing doses. Clear water is a look, not a test result; only a liquid kit sees new tank syndrome.

How long does it take to cycle a new tank?

About 4–6 weeks fishless, longer if cold. "Cycled" means 2 ppm ammonia converts to 0 ammonia and 0 nitrite within 24 hours while nitrate rises.

Fish-in or fishless cycle?

Fishless — you grow the bacteria with zero animals at risk. Fish-in works only as damage control: daily testing, daily changes, hardy species, light feeding, and luck. If fish are already in the tank, run the emergency protocol above instead.

Fish are dying right now — what's the very first step?

A 40–50% temperature-matched, dechlorinated water change, then test ammonia and nitrite. Repeat the change daily until both hold at 0 ppm, no feeding meanwhile. The checklist PDF walks the whole sequence.

Can new tank syndrome kill fish?

Yes — the ammonia and nitrite spikes of an uncycled tank are the #1 killer of new-tank livestock, invisible and odorless at killing doses, with deaths clustering in weeks 1–6. Any death in a tank under six weeks old is new tank syndrome until a liquid test proves otherwise.

How long until a new tank is safe for fish?

4–6 weeks fishless (6–8 if cold), and the test — not the calendar — is the green light: 2 ppm ammonia must convert to 0 ammonia and 0 nitrite within 24 hours while nitrate rises. Seeding with media from an established tank can cut it to 2–3 weeks.