Water changes: what the science actually says about frequency, volume, and the fear of killing fish

Few topics in fishkeeping generate as much confident, contradictory advice as water changes. You will find experienced keepers who have not changed their water in years and have healthy, thriving tanks. You will find equally experienced keepers who change 50% twice a week and attribute their fish's longevity to exactly that. You will find people who insist a large water change killed their fish, and people who have done 90% changes for a decade without losing a single animal.

The disagreement exists largely because the underlying mechanisms are rarely discussed. Once you understand why water changes help — and what actually causes the rare deaths that get attributed to them — most of the debate resolves.

The four mechanisms

A water change does four things, and they are not equally important in every tank:

1. It dilutes dissolved waste products. Nitrate, phosphate, dissolved organic compounds, hormones released by fish, and any other substances that accumulate in a closed water system are reduced in proportion to the volume exchanged. A 50% water change with clean replacement water halves their concentration.
2. It replenishes minerals. Most tap water contains calcium, magnesium, and carbonate hardness (KH). In a closed aquarium, KH in particular is slowly consumed by the acidifying processes of nitrification and organic decomposition. Water changes restore it.
3. It stabilises pH. Directly linked to KH replenishment. As KH depletes, the water loses its buffering capacity and pH becomes unstable — a particular problem in high-CO₂ planted tanks and tanks with aquasoil substrates. Regular water changes maintain the buffer that keeps pH steady.
4. It removes dissolved organics that cannot be filtered. Mechanical and biological filtration removes solid waste and processes ammonia and nitrite into nitrate. It does not remove the dissolved organic compounds — tannins, proteins, breakdown products — that accumulate over time and give old tank water its characteristic yellow tint.

Notice that beneficial bacteria are not on that list. Biological filtration lives in the filter media and substrate — not in the water column. Water changes do not meaningfully disturb your cycle.

What nitrate actually does, and at what levels

Nitrate is the end product of the nitrogen cycle. Ammonia from fish waste is processed by bacteria into nitrite, and then into nitrate. Nitrate does not accumulate to dangerous levels quickly — it is far less acutely toxic than ammonia or nitrite — but it does accumulate continuously in any tank with fish, and it matters at chronically elevated concentrations. The comammox guide covers the biology of this conversion in detail.

Aquatic toxicology research on freshwater fish consistently finds that nitrate toxicity thresholds vary significantly by species. Most commonly kept tropical community fish — tetras, rasboras, livebearers, corydoras — tolerate moderate nitrate well, with acute toxicity typically occurring at concentrations well above what most hobbyist tanks reach. However, chronic exposure at lower levels has been shown in laboratory studies to affect growth rates, reproductive success, immune function, and long-term organ health — particularly in the liver and kidney.

The species sensitivity gradient is steep. Discus and other cichlids from soft, acidic blackwater environments are notably sensitive, and many experienced discus keepers target nitrate below 20 ppm through frequent large changes. Goldfish, conversely, tolerate comparatively high nitrate. Most community fish sit somewhere between these extremes. A working practical target for most tropical setups is to keep nitrate below 40 ppm, though lower is generally better and some sources suggest that below 20 ppm is preferable for long-term fish health.

One practical complication for UK keepers: tap water already contains nitrate, sometimes significantly. UK water companies are legally required to keep tap water below 50 mg/L nitrate (roughly 50 ppm NO₃), but supplies near agricultural areas routinely contain 20–40 ppm. In those areas, a water change with untreated tap water raises the floor: you are replacing tank water with water that already has substantial nitrate in it. If this applies to you, your water company publishes nitrate data for your supply — it is worth checking. RO water eliminates this problem entirely.

Dissolved organics and what we do and do not know

Nitrate is measurable and well-studied. The broader category of dissolved organic compounds (DOC) is considerably less well-characterised in an aquarium context, and honesty requires acknowledging that.

What we know: DOC accumulates in closed water systems. It includes proteins, amino acids, humic acids, phenolic compounds, and the breakdown products of uneaten food and decomposing plant matter. At high concentrations, some of these compounds are harmful to fish — affecting gill function, oxygen uptake and general physiological stress. The yellowing of long-unchanged tank water is largely DOC accumulation, and the smell of old tank water reflects the same.

What we are less certain about: the precise threshold at which DOC accumulation begins to harm fish in a typical well-maintained tank, whether different compounds in DOC have meaningfully different effects, and whether fish hormones and stress compounds released into the water column have measurable impacts at normal tank concentrations. Some aquaculture research suggests that chronic exposure to conspecific stress hormones can suppress immune function, but the evidence base for this in typical hobbyist conditions is not strong. We should be honest that this is an area where the data is limited.

What we can say: DOC accumulation is real, is not addressed by filtration alone, and is reduced by water changes. It is a reason to change water even in tanks where nitrate is being handled by plants.

Why fish sometimes die after large water changes — and why it is almost never the volume

This is the claim the hobby repeats most confidently and understands least clearly: large water changes kill fish. Some experienced keepers refuse to change more than 10–15% at a time because of it.

The evidence does not support volume as the cause. What the evidence does support is that fish sometimes die after water changes, and the mechanisms are well-understood — none of which involve the percentage of water changed:

1. Old tank syndrome — the most common cause by far

Old tank syndrome is real, documented, and the likely explanation behind the majority of post-water-change deaths attributed to "doing too much."

Here is the sequence: a tank receives no or minimal water changes over months or years. The biological processes of nitrification — the conversion of ammonia to nitrite to nitrate — are mildly acidic. Organic decomposition adds CO₂ and organic acids. Slowly and continuously, these processes consume carbonate hardness (KH). As KH depletes, buffering capacity is lost, and pH begins to drop. The fish adapt gradually — they may have been living in water at pH 5.5–6.0 for months, having adjusted their physiology to this increasingly hostile environment. The keeper, seeing fish that look broadly healthy, may not realise anything is wrong.

Then a large water change is made with tap water at pH 7.5–8.0 and 5–6 dKH. In a 50% change, the tank pH swings upward by one to two units in the space of minutes. For fish whose physiology has adapted to strongly acidic water, this is a severe shock. They die. The keeper concludes that the large water change was the problem.

The actual problem was months of neglect and water degradation. The water change was the trigger for the death, not its cause — much like removing a support from a structure that was already failing. Had regular water changes been made all along, there would have been no dangerous parameter gap between tank and tap water to cross.

2. Temperature mismatch

Freshwater fish are sensitive to rapid temperature change. Introducing large volumes of water several degrees colder or warmer than the tank can cause thermal shock, suppress immune function, and trigger outbreaks of opportunistic pathogens within 24–48 hours. The fish do not always die immediately — they develop symptoms over a day or two, by which point the water change has been forgotten as a cause. Temperature-matching replacement water to within 1–2°C of the tank is simple and eliminates this risk entirely.

3. Untreated chlorine and chloramines

The tap water used in most water changes contains chlorine or chloramines added by the water company. Chlorine off-gasses in minutes to hours; chloramines do not — they are stable and require a dechlorinator specifically rated for chloramine removal. Gill damage from chloramine exposure can be severe and is not always immediately fatal — fish may appear stressed and decline over 12–24 hours, which again distances the cause from the symptom in the keeper's mind. The chlorine and chloramines guide covers this in detail. Dechlorinating before adding replacement water removes this risk.

What this means for large changes

Address those three things — match temperature, dechlorinate, maintain consistent water chemistry through regular changes — and the volume becomes essentially irrelevant within normal bounds. Fish breeders, koi keepers, and aquaculture operations routinely perform 50–90% daily changes on valuable and sensitive animals without harm. The planted aquarium competition world does the same. The fear of large water changes is a fear of a poorly understood mechanism, not of volume itself.

The planted tank exception — and its limits

A densely planted, well-lit, CO₂-supplemented tank running fast-growing stem plants can consume nitrate faster than fish produce it. In this specific scenario, nitrate may genuinely stay near zero without any water changes, and keepers have maintained healthy tanks this way for years. Diana Walstad's work on the planted tank method — which uses a soil substrate and relies heavily on plant nutrient uptake and substrate microbial activity — demonstrated this convincingly.

But the planted tank exception is narrower than it is often presented:

• It addresses nitrate — not DOC. Plants consume nitrate. They do not remove dissolved organic compounds, fish hormones, or the accumulated products of organic decomposition. These still build up in a no-change planted tank.
• It depends on specific conditions. The plant mass, light level, CO₂ availability, stocking level and feeding regime all have to be in balance. A planted tank with moderate light and low-growing plants is not consuming nitrate fast enough to offset a meaningful fish load.
• KH depletion continues. Plants do not replenish carbonate hardness. A planted tank on a CO₂ system with no water changes will see progressive KH depletion — exactly the conditions for old tank syndrome — unless the substrate or substrate additives buffer the water. Aquasoil actively depletes KH. CO₂ stability and KH are tightly coupled, and both are affected by water changes (or the absence of them).

For most tanks — any stocked tank without fast-growing plants under strong light — regular water changes are not optional. They are the primary mechanism for maintaining water quality in a closed system.

10% weekly? 50% weekly? Monthly large changes? What does the evidence support?

There is no single universally correct answer here, and the research base for specific percentage and frequency recommendations in hobbyist settings is thinner than the confidence with which advice is given might suggest. What the evidence does support:

Frequent smaller changes outperform infrequent large changes for parameter stability. A 25% change every week keeps nitrate, DOC, and parameter variation more stable across time than a 90% change once a month — even though the cumulative water volume exchanged is similar. The reason is that in the monthly-change scenario, parameters are allowed to drift significantly for three weeks before being reset. Fish live in the degraded conditions for most of their lives.

Volume matters more than most keepers allow. The dilution maths are unforgiving. At the common recommendation of "10–15% weekly," a moderately stocked tank producing 20 ppm of nitrate per week reaches a steady-state nitrate of 130–200 ppm — well above any sensible target. The "do a little every week" advice is not wrong in principle but is frequently wrong in the quantities actually recommended.

Stocking level should drive the percentage. A lightly stocked planted tank may genuinely be fine with 20–25% weekly. A heavily stocked community tank, a cichlid setup, or any breeding tank typically needs 40–50% or more. Goldfish, which produce exceptional waste relative to their size, routinely need 50–75% weekly. There is no substitute for measuring nitrate and working backward to find the change volume that keeps it where you want it.

The upper limit of safe volume is much higher than the hobby assumes. With properly temperature-matched, dechlorinated water and a well-maintained tank, 80–90% changes are physiologically safe. Whether they are necessary is a separate question — but the fear of them is not evidence-based.

Honest gaps in the science

Much of what gets stated confidently about water changes in the hobby is extrapolated from general principles and practical observation rather than controlled aquarium-specific research. Some genuine uncertainties:

• Specific DOC thresholds. We know dissolved organics accumulate and that high concentrations are harmful. We do not have robust data on the specific compound concentrations at which harm begins in typical hobbyist conditions.
• Whether fish benefit from water change "stress." There is a fringe argument that the minor osmotic challenge of a water change is itself beneficial — triggering a mild stress response that improves immune readiness, analogous to the mild stress of exercise. The evidence for this in fish is essentially non-existent, but it is occasionally raised as a reason to prefer larger, less frequent changes.
• The role of dissolved hormones. Aquaculture research documents that fish release stress hormones and alarm substances into the water and that conspecifics respond to these chemically. Whether accumulation of these substances in a home aquarium reaches concentrations that meaningfully affect fish behaviour or physiology is not established with confidence. It is plausible but not proven.
• Optimal parameters for long-term versus short-term health in specific species. We have reasonable data on acute nitrate toxicity. We have much less data on the chronic effects of low-level elevations over a fish's full lifespan — the kind of data that would tell you definitively whether 30 ppm versus 15 ppm makes a measurable difference in a three-year-old neon tetra.

None of these uncertainties undermine the case for regular water changes. They simply mean we should hold some of the more specific numerical recommendations with appropriate tentativeness.

What the evidence actually supports

Water changes work through dilution of waste products, replenishment of KH and minerals, and removal of dissolved organics that filtration cannot address. Nitrate is the most measurable target, but not the only one. The maths of dilution mean that volume matters — small weekly changes in heavily stocked tanks cannot achieve low nitrate regardless of how consistently they are done.

Fish deaths attributed to large water changes are almost universally caused by old tank syndrome (parameter gap from neglect), temperature mismatch, or untreated chlorine — not by the volume of water exchanged. A well-maintained tank with consistently replaced, temperature-matched, dechlorinated water can safely receive changes of 50–90% without harm. The fear of large changes is not evidence-based.

Planted tanks with fast-growing plants under strong light and CO₂ can hold nitrate near zero, but still accumulate dissolved organics and experience KH depletion — particularly on CO₂ systems. The planted tank is not exempt from water changes; it changes the maths on nitrate specifically, not on overall water quality.

Where the science is less clear — on specific DOC thresholds, the effects of dissolved hormones, and optimal chronic nitrate levels for specific species — the honest position is that we do not know precisely, and the safe choice is to err toward more exchange rather than less.