How Long Can Betta Fish Live Without Eating — and What to Do With the Answer

A healthy adult betta can go roughly 10 to 14 days without eating. That ceiling comes from keeper practice rather than from a published trial, because no controlled feed-deprivation study of Betta splendens has been run to mortality. The measured evidence covers a much shorter window. In a 2025 study in the journal Animals, male bettas averaging 1.56 g lost 15.0 ± 1.5 percent of their body weight after about 60 hours without food during overland shipping; every fish survived, and the group was back above its starting weight eight days after feeding resumed. In a closed aquarium the water fails before the fish does, so the reading that decides the outcome is ammonia, converted against your pH and temperature.

How long can a betta fish live without eating?

For eleven years my job has been to follow a calorie figure backwards through a frozen-food plant, from the carton to the costing sheet to the moisture-loss table that changed the denominator in the middle. That work leaves one reflex behind: before repeating a number, find out what sits under it.

Under the 10-to-14-day betta figure there is nothing. No starvation trial, no survival curve, no sample size. It circulates because it roughly matches what keepers observe and nobody has reason to run the experiment that would test it.

What exists is a fasting measurement taken for another purpose. Thongprajukaew and colleagues, publishing in Animals in 2025, fasted male bettas for 12 hours, shipped them overland for about 48 hours, and weighed them on arrival. Mean loss was 15.0 ± 1.5 percent of body weight in fish that started at 1.56 ± 0.02 g. Survival was 100 percent. Fed again, the same fish read −3.36 percent at day eight and −6.99 percent at day twelve, having overshot their original weight.

Fifteen percent in two and a half days sounds alarming until you open the denominator. That figure bundles an emptied gut, handling stress and osmotic loss in a bag, with the fish held at 31.1 ± 0.3 °C, near the top of the species' range. It is the total mass change of a shipped fish, an upper bound on what a short fast costs, and not a daily burn rate to multiply out.

Age changes the answer as much as duration does. James and Sampath, in a 2004 feeding-frequency study, raised 30-day-old betta juveniles for 77 days on schedules ranging from one meal every three days to three meals daily. The once-every-three-days group ate 266 mg of dry feed and converted it to 61 mg of flesh. The once-daily group ate 543 mg and put on 175 mg. Those fish still grew and still spawned. What deprivation cost them was scale, 274 eggs against 1,833 for daily feeding.

An adult betta's survival window is long enough that it almost never becomes the binding constraint. A young betta's window for growing normally is far shorter than its window for staying alive.

Why the water gives out before the fish does

An unfed betta is a slower ammonia source. UF/IFAS Extension's fact sheet FA16, Ammonia in Aquatic Systems, puts it plainly: the more feed a fish receives, the more ammonia it produces, though even a starved fish produces some. A fast lowers the nitrogen load on a small tank. The harm comes from what people add to prevent the fast, since uneaten feed decays into the same ammonia and a five-gallon tank has no dilution to absorb it.

That same fact sheet supplies the sentence I wish every keeper read before searching for a duration: fish are not likely to eat during periods of ammonia stress. The Merck Veterinary Manual arrives there clinically too, calling poor water quality the most common cause of environmentally induced disease in fish and noting that fish show very few specific signs of illness. Appetite loss is among the least specific of all, appearing in ammonia and nitrite exposure, temperature crashes, velvet and bacterial infection alike. The search that brought you here often has the arrow backwards.

Why two official ammonia numbers disagree, and which one applies to your tank

Two published figures appear to contradict each other.

The US EPA, in its 2013 freshwater ammonia criteria (fact sheet 820-F-13-013), recommends an acute limit of 17 mg total ammonia nitrogen per liter at pH 7 and 20 °C as a one-hour average, and a chronic limit of 1.9 mg TAN/L as a 30-day rolling average, with no four-day average above 4.8.

UF/IFAS says that anytime un-ionized ammonia exceeds 0.05 mg/L the fish are being damaged, and that at 2.0 mg/L they die.

One appears to tolerate 1.9, the other calls 0.05 harmful. The gap is a denominator problem, the same one that puts two different calorie counts on two honest labels for the same chicken breast: raw weight against cooked.

EPA reports total ammonia nitrogen, everything the test kit detects. UF/IFAS reports only the un-ionized fraction, NH₃, which crosses the gill and is around 100 times more toxic than the ionized form. Converting between them requires pH and temperature, using the equilibrium relationship from Emerson, Russo, Lund and Thurston's 1975 paper in the Journal of the Fisheries Research Board of Canada, the same table reproduced in the back of the UF/IFAS sheet.

Run EPA's 1.9 mg TAN/L through that relationship at pH 7 and 20 °C and it becomes 0.0075 mg/L of un-ionized ammonia. Measured against the 0.05 mg/L damage line, EPA's chronic criterion turns out to be roughly 6.7 times stricter, because it was written to protect unionid mussels and gill-breathing snails, which are more sensitive than fish. No contradiction. Different units, different animals.

The practical consequence is that your test-card number means nothing until you attach a pH to it. These are calculated values rather than measurements, the Emerson relationship applied to one reading at 26 °C:

| Test-kit reading (total ammonia nitrogen) | Tank pH at 26 °C | Un-ionized ammonia nitrogen | Share of the 0.05 mg/L damage line | |---|---|---|---| | 0.5 mg/L | 7.0 | 0.003 mg/L | 6% | | 0.5 mg/L | 7.5 | 0.009 mg/L | 19% | | 0.5 mg/L | 8.0 | 0.029 mg/L | 57% | | 0.5 mg/L | 8.5 | 0.081 mg/L | 162% |

The same 0.5 reading is 27 times more toxic at the bottom of that table than at the top.

I learned that conversion the expensive way, in my second year of keeping bettas, on hard well water. I read 0.5 mg/L, checked it against the 0.05 in my notebook, decided I had ten times the margin I needed, and left it. My pH was 8.5. The un-ionized concentration sat above the damage line rather than a tenth of it, because I had compared a total against a fraction, precisely the error I get paid to catch on a costing sheet. The fish declined over a week and did not recover. FishBase gives Betta splendens a pH range of 6.0 to 8.0, so that tank was outside the species' envelope before ammonia entered the arithmetic.

Nitrite is the reading most keepers skip

Nitrite sits between ammonia and nitrate in the filter's chain, and it is the reading people forget once a tank has settled. Claude Boyd, writing for the Global Seafood Alliance in January 2014, reports that aquaculture systems usually hold nitrite below 0.1 mg/L, and that 0.5 to 2.5 mg/L nitrite-nitrogen is the band in which warmwater species avoid growth reduction and increased disease susceptibility. Coldwater fish are more sensitive, their equivalent band starting at 0.0125 mg/L.

The mechanism explains why an affected fish stops feeding. Nitrite crosses the gill and converts hemoglobin to methemoglobin, which cannot carry oxygen. In channel catfish exposed for 24 hours, Boyd's 1990 data give 21 percent methemoglobin at 1 mg NO₂-N/L and 60 percent at 2.5 mg/L. Respiratory stress appears above 5 percent, clear toxicity at 20 to 30 percent, mortality above 50 percent. A fish suffocating internally has no appetite.

Dissolved oxygen decides how much of this a fish tolerates. Jesse Chappell of Auburn University Extension puts it as a comparison: a moderately affected fish at 7 ppm dissolved oxygen should survive, the same fish at 2.0 ppm likely will not. Chloride competes with nitrite at the gill, which is why he recommends holding a chloride-to-nitrite ratio of at least 9 to 1.

One limit I should state rather than paper over. I have never measured methemoglobin in a betta and know of no published measurement in the species. Those percentages are channel catfish figures from ponds, carried here across a species boundary and a system boundary. What I can vouch for is the arithmetic and the direction: nitrite in a betta tank should read zero on a hobby kit, and any detectable reading means the filter is not keeping up.

Temperature moves both numbers at once

FishBase lists Betta splendens at 24 to 30 °C. Field measurements from five marshland habitats in Chiang Rai province, northern Thailand, published by Ramos, Gonçalves and Vasconcelos in PeerJ in 2025, recorded 22.1 to 25.1 °C, pH 4.8 to 5.8, water 0.16 to 1.20 m deep, and dissolved oxygen between 0.19 and 3.02 mg/L. Wild bettas manage that acidic, oxygen-poor water by breathing air through the labyrinth organ.

Temperature works on both sides of the problem. Warmer water raises metabolic demand, so reserves deplete faster during a fast, and it raises the un-ionized fraction of whatever ammonia is present: at pH 7.5, the Emerson relationship gives 1.64 percent at 24 °C against 2.16 percent at 28 °C. A heater failure that drops a tank to 18 °C and one that pushes it to 32 °C both produce a fish that stops eating, and only a thermometer tells you which you have.

A weekend away, a week away, and a fish that has stopped eating

The same duration question serves very different decisions, and separating them changes what you should do.

| | Weekend absence (2-3 days) | Longer absence (7-14 days) | Unexplained refusal | |---|---|---|---| | What it is | Planned fast | Absence beyond a safe fast | A symptom | | Expected mass change | Slight, recovered in days | Measurable; 15 percent is the ceiling | Progressive if it continues | | Main water risk | Food added in your absence | Filter or heater failure, unwatched | Ammonia or nitrite already high | | First move | Feed normally, leave it alone | A person, not a device | Test ammonia, nitrite, pH, temperature | | Escalate when | It does not | Day 8 without a check | Refusal passes 4-5 days |

The strongest argument against my position deserves stating in full, because I have seen it used well. Fasting is not free. James and Sampath's juveniles show that repeated deprivation carries a real cost even when the fish grow and spawn anyway: 61 mg of flesh against 175, and 274 eggs against 1,833. "They can go two weeks" has become a license for chronic underfeeding, and a keeper invoking it to excuse a thin, stunted young fish is misusing a number that describes an emergency ceiling.

All of that is true, and none of it changes what to do about a three-day gap. That study measured a schedule sustained across 77 days in growing juveniles, not one interruption in an adult. A short fast is close to free and a repeated one is not, which is why absences past a week call for a person rather than a device.

What to do once you have the figure

A number earns its keep where it changes an action. These steps are ordered by what fails first.

1. Test before you change anything. Ammonia, nitrite, pH and temperature, from the tank and not from the tap. A water change performed first destroys the evidence.
2. Convert the ammonia reading. Multiply total ammonia nitrogen by the un-ionized fraction for your pH and temperature, using the table in UF/IFAS FA16. Compare the product against 0.05 mg/L, never the raw reading.
3. Change water if either nitrogen number is elevated. FA16 recommends 25 to 50 percent. Match the temperature and dechlorinate.
4. Correct temperature into 24-27 °C before drawing any conclusion about appetite. A cold betta that resumes eating at 26 °C was never a feeding problem.
5. Stop feeding while the water is out of range. FA16 is explicit that reducing or eliminating feeding is the first response to ammonia, since uneaten feed makes it worse and a stressed fish will not eat anyway.
6. Re-offer one small portion daily and write down whether it was taken. A record separates a fish that ate on day three from one you assume did.
7. Escalate at four to five days of total refusal, or immediately on any urgent sign below.

Until about 2021 I told people to leave a plaster vacation block in the tank over a long weekend. I stopped after working the arithmetic the way I would work a yield sheet. The block adds mass to a volume with no dilution and no observer, while the fast it prevents costs a healthy adult almost nothing. The safer option was the one I had been arguing against. For anything past a week I now give the same answer: a neighbor with a container of pre-measured portions.

Questions keepers actually ask

How do I know if my betta fish is starving?

Starvation shows in body shape before behavior. Look for a sunken belly, a spine ridge along the back, and a head that appears large against a narrowing body. A betta refusing food while still flaring and patrolling is likelier reacting to water quality. Test the water first.

Can a betta fish go a weekend without food?

Yes. A healthy adult betta handles two or three days unfed with no lasting effect, and fasting lowers the ammonia it produces rather than raising it. The risk comes from food left behind to dissolve. Skip the vacation block, change water before leaving, feed normally on your return.

Can I not feed my betta for 2 days?

Two days without food is routine and safe for a healthy adult betta, and some keepers fast deliberately one day per week. Do not compensate with a double portion before or after. Uneaten feed decays into the ammonia that causes the actual harm.

Is it normal for a betta fish to not eat for 2 weeks?

No. Two weeks of refusal is a symptom rather than a fast. UF/IFAS notes that fish are unlikely to eat during periods of ammonia stress, so test ammonia, nitrite, pH and temperature first. Refusal past four or five days warrants investigation rather than waiting.

How long can betta fish live without eating in the wild?

No field study measures fasting duration in wild Betta splendens, and the species' natural history is thinly researched. Habitat work from Chiang Rai, Thailand recorded shallow marsh pools 0.16 to 1.20 m deep with dissolved oxygen as low as 0.19 mg/L. Wild fish forage continuously there rather than fasting.

Which signs make appetite loss urgent?

Refusal combined with any of these needs same-day action: rapid or labored gill movement, gasping at the surface, clamped fins, resting on the substrate, a detectable ammonia or nitrite reading, or visible weight loss along the spine. Refusal alone in an alert fish can wait one feeding.