Few habits in home growing are as widespread, or as argued over, as flushing. In the last week or two before harvest, many growers stop feeding and give their plants plain water. Ask why, and the answers come quickly: the smoke will be smoother, it will taste cleaner, it will "burn clean" and leave white ash rather than black. Ask for the evidence, and the conversation usually gets shorter.

CannaTinker's grow guide, built from UK small-grow practice, includes a flush. When the trichomes are about 90 % milky and the first amber heads appear, feeding stops. For the last 7–10 days the plants get clean water adjusted to pH 6.0, with a generous runoff, so that the EC of the water draining from the coco falls from about 2.0 to below 0.6 mS/cm. The tutorial is careful about what this does: water rinses excess salts out of the growing medium, not out of leaves and flowers.

Until recently there was almost no research to set beside this practice. That has changed. Two university groups have now run controlled flushing trials, and their results are clearer than most of the online debate. They are also narrower than they first appear: high-CBD hemp, greenhouses and growth chambers, and no tasting at all. This article goes through what they did, what they found, and what a careful small grower can take from it. CannaTinker's position, explained at the end, is that flushing remains a reasonable small-grow practice when it is done briefly and with moderation.

A note before we start: in the UK, growing cannabis, even with a prescription, requires a Home Office licence. This article is educational and describes published research.

What flushing is, and what it is not

The word covers two different things, and keeping them apart clears up a lot of confusion.

The first is leaching: running enough water through the pot to wash out dissolved fertiliser salts that have built up in the growing medium. Growers track this with an EC meter. EC, or electrical conductivity, measures how much dissolved salt is in a solution. Fertiliser raises it, and plain water lowers it. Comparing the EC of the feed going in with the runoff coming out shows whether salts are accumulating. CannaTinker's tutorial warns that a runoff EC 0.6 above the input means salts are building up.

The second is nutrient deprivation: once the medium has been rinsed, the plant gets no fresh fertiliser for the rest of its life. This is the part growers believe changes the flower.

Hershkowitz (2024) describes the typical sequence: the medium is irrigated with pure water until the runoff EC is close to that of the irrigation water, sometimes with commercial "finishing products" added, and the plant then gets tap water or a dilute solution until harvest. Alden and Faust (2025) define flushing as stopping fertiliser and then leaching the substrate, followed by clear water during the final 1 to 3 weeks before harvest.

What flushing cannot do is wash anything out of a leaf or a bud. Water poured on the medium reaches the roots, not the stored contents of the plant. Any change in the flower's chemistry has to come from the plant itself, by using up or moving what it already holds.

The claims: taste, a "clean burn" and white ash

The claims for flushing are almost entirely about the experience of smoking. Hershkowitz (2024) summarises the anecdotal case: flushing is said to improve flavour, throat feel and combustion by reducing the concentrations of chlorophyll and nutrients in the flower. Consumers often use the colour of smoke or ash as a quality signal, with light or white ash taken to mean better flower.

That last claim has an alternative explanation. As Hershkowitz reports, the trade writers Justice and Roggen, who described the white-ash belief, themselves suggested that ash and smoke colour depend on combustion temperature or water content. Hershkowitz adds a comparison with tobacco, where moisture content, curing, blend and packing density all affect how a cigarette burns. In other words, ash colour may say more about how the flower was dried and cured, and how it was smoked, than about what it was fed.

None of this proves the claims wrong. It shows that they have not been tested. Alden and Faust (2025) put it plainly: no study to date has shown that flushing improves smoking quality, so the evidence for that purpose is anecdotal. Notably, they do not dismiss the possibility. They write that they suspect the sensory character of dried flowers "might differ" when plants have been under prolonged nutrient stress compared with well-fed plants. Nobody has yet done the tasting to find out.

Where flushing may have come from

The honest answer is that nobody knows. Alden and Faust (2025) say directly that "the origin of flushing is unknown", but they offer a plausible hypothesis.

Salts accumulate in the growing medium when a plant is fed more than it uses. High salt levels around the roots cause salt stress: stunted growth, dead tissue at leaf edges, wilting. In commercial growing of ornamental plants, leaching with clear water to bring the medium's EC down is a standard correction for exactly this imbalance. Alden and Faust suspect that cannabis growers, often feeding heavily with expensive specialist fertilisers, adopted the same remedy. Somewhere along the way it acquired a new name, "flushing", and a new reputation as a way to improve the final product. The recovery of plants once salt stress was relieved may later have been read as an improvement in quality.

The authors note that overfeeding is common in the cannabis industry. In one survey they cite, half of growers spent 10–29 % of their production budget on fertiliser, against under 2 % in ornamental greenhouse growing. If feeding matches the plant's needs, they argue, flushing is not needed to avoid salt stress.

This is a hypothesis, not a finding. But it fits the tutorial's own framing: the first job of a flush in coco is to bring a salty medium back towards neutral.

What the trials measured and found

Clemson: two hemp cultivars in a greenhouse

The most detailed study is by Michael Alden and James Faust at Clemson University in South Carolina (Alden and Faust, 2025). They worked with two high-CBD cultivars, 'THM Jack' and 'Southern OG', both chemotype III, which means CBD-dominant with less than 0.3 % THC. Rooted cuttings grew in 2.5-litre pots of a peat-based mix, eight plants per square metre, in a greenhouse with blackout curtains to give a 12-hour day. Sunlight was topped up by LEDs, giving an average daily light integral of about 39 mol·m⁻²·d⁻¹, and the average temperature was about 23 °C. Plants were fed continuously at 200 mg/L nitrogen. The experiment was run twice, with four plants per treatment in each run.

The flush was thorough. Pots were drip-irrigated with tap water (EC about 0.1 mS/cm) until the runoff measured 0.2 mS/cm or less, and afterwards received only tap water. Getting a 2.5-litre pot down to that level took about 15 litres of water. 'THM Jack' was flushed for 0, 1, 2 or 3 weeks before harvest at seven weeks of flower; 'Southern OG' for 0 to 4 weeks before harvest at eight weeks. The researchers measured dry mass, cannabinoids and mineral nutrients in the flower. They did not test taste or smoke.

The results, briefly:

  • Flower mass fell as flushes got longer. In 'THM Jack', trimmed flower averaged 0.17 kg/m² without a flush and lost about 0.01 kg/m² for every week of flushing. In 'Southern OG', a two-week flush cut flower mass by 10 % (from 0.40 to 0.36 kg/m²), and going from two to four weeks cost a further 22 %. Total reproductive mass in 'Southern OG' was identical after a one-week flush and with no flush, so short flushes cost least.
  • Potency barely moved. THC concentration, very low in these hemp plants, did not change. CBD concentration did not change in 'THM Jack', but rose by about 0.5 percentage points per week of flushing in 'Southern OG', from 12.8 % in the control. Because flower mass fell faster, CBD per square metre declined in both cultivars.
  • Nutrients in the flower fell. Nitrogen, phosphorus and potassium concentrations were highest in unflushed plants and declined with each week of flushing. Phosphorus, for example, dropped by about 0.09 percentage points a week. Calcium rose in 'Southern OG', and magnesium did not change.
  • Leaves showed classic deficiency. Older leaves yellowed, the yellowing progressed to dead edges, and leaves dropped. In 'THM Jack', leaves often turned orange and purple first.

Alden and Faust also listed possible side benefits, and labelled them speculative. Losing lower leaves might shed foliage often infected with powdery mildew and reduce the moisture load in drying rooms. The higher calcium in 'Southern OG' flowers might, by analogy with petunia and rose, lower susceptibility to grey mould (Botrytis). A small rise in sulfur raised the question of whether sulfur-containing aroma compounds change. None of these was tested.

Their commercial conclusion was blunt. Using 'Southern OG' as a model, they estimated that a one-week flush would cost about $39 per square metre in lost revenue while saving about $0.13 in fertiliser. They concluded that thorough leaching protocols like theirs should not be used in commercial production without a specific goal.

The same experiment, published twice

Alden's doctoral dissertation (Alden, 2025) includes a chapter with the same title and the same data. It is useful for detail, but it is not a second, independent trial. Counting it twice would make the evidence look stronger than it is. The dissertation's summary says that flushing "tended to decrease" dry mass, cannabinoid yield and mineral nutrient concentrations, while noting that these changes are thought to improve the sensory character of dried flowers.

Utah: a growth chamber and a gentler alternative

The second group is at Utah State University. In her master's thesis, Julie Hershkowitz (2024) devoted a chapter to flushing. Its title summarises the finding: preharvest flushing "does not reduce mobile nutrient concentrations in flower, but may reduce yield".

Conditions were quite different from Clemson's. Two high-CBD cultivars, 'Cherry' and 'T1', grew in 6.7-litre pots of a peat-based mix in a walk-in growth chamber. Full-spectrum LEDs gave about 970 µmol·m⁻²·s⁻¹ on a 12-hour day, a daily light integral of about 42. The temperature was about 26 °C by day, and CO₂ was enriched to 1,200 ppm, roughly three times outdoor levels. All plants were fed for the first four weeks of flower and harvested at 56 days. Each experiment had three or four replicates per treatment.

Two kinds of flush were compared. In the first experiment, pots were leached with reverse-osmosis water to a runoff EC below 0.5 mS/cm and then fed a complete solution with the nitrogen left out, for up to 21 days. In the second, pots were leached with tap water to below 0.5 and then given only tap water for 7 or 14 days.

The results differed by method. Withholding only nitrogen did not significantly change dry flower yield, which averaged 571 g/m² across treatments. Plain tap water did reduce yield, and the longer the flush, the bigger the loss. Bud diameter was not significantly affected. CBD concentration was unchanged by the nitrogen-free feed and rose slightly with tap water, echoing the 'Southern OG' result. CBD yield per plant did not differ significantly between treatments. In both experiments, leaves yellowed and chlorophyll fell as flushes lengthened.

The most interesting finding concerns nutrients. Nitrogen in the leaves fell during flushing, but nitrogen in the flowers stayed roughly steady. Hershkowitz reports similar patterns for phosphorus, potassium, calcium, sulfur and magnesium, with no significant difference in flower concentrations of mobile nutrients between flushed and unflushed plants.

Earlier studies and why they disagree

Both groups review earlier, less detailed work. As Alden and Faust (2025) describe them, studies by Stemeroff (2017) and Wedryk (2019) found that flushing for two weeks or less did not affect flower yield or nutrient and cannabinoid concentrations, and a study by Saloner and colleagues (2024) found that a two-week flush did not affect yield. Alden and Faust offer an explanation. The earlier studies did not report runoff measurements, and Saloner's runoff was still 0.37 mS/cm three days into the flush. The medium may still have held enough nutrients to keep the plants growing, so the "flushed" plants were not truly deprived for the whole period. Hershkowitz (2024) makes a similar point, citing work on poinsettia in which the growing medium held enough nutrients for growth up to 42 days after feeding stopped.

So the trials do not show that every flush costs yield. They show that a flush thorough enough to cut off nutrients completely does, and that the cost grows with time.

What the plant is doing: remobilisation

Why would a plant keep its flowers supplied when the roots find nothing? The answer is a normal part of plant life, called remobilisation.

Some nutrients, among them nitrogen, phosphorus, potassium and magnesium, are mobile inside the plant. When supply from the roots falls short, or as leaves age, the plant breaks down proteins and chlorophyll in its older leaves and ships the nutrients through the phloem (the plant's sugar-transport tissue) to the parts that are still growing. Flowers are strong "sinks" that draw resources towards themselves. Hershkowitz (2024) cites figures from other crops: maize remobilises about 40 % of its leaf nitrogen, and wheat up to 90 %, to feed developing seeds. A poor external supply makes leaves age and give up their nutrients sooner.

This is exactly what the yellowing fan leaves of a flushed plant show. The tutorial describes it as the leaves turning "autumn colours" as the plant uses up its own nitrogen reserves, and the trials agree. Hershkowitz's data, with leaf nitrogen falling and flower nitrogen holding steady, are what remobilisation predicts: the plant sacrifices old leaves to protect the flowers.

The two studies disagree on whether flower nutrients fall. At Clemson, nitrogen, phosphorus and potassium in the flower declined with each week of flushing. At Utah, flower concentrations held steady. Several differences could explain this: smaller pots at Clemson (2.5 versus 6.7 litres), a stricter leaching target (0.2 versus 0.5), longer flushes in one cultivar, different cultivars, and very different light and CO₂. The point that matters for growers is that even where flower nutrients fell, nobody measured whether anyone could taste the difference.

Why the trials may not fully transfer to a tent

These are careful studies, but they answer a commercial grower's question about a particular kind of crop. Several differences matter for a home tent.

  • The plants. All four cultivars were high-CBD types, and the Clemson pair were legally hemp, with under 0.3 % THC. Responses to nutrient stress may differ in THC-dominant cultivars, and Alden and Faust themselves found that the two cultivars they tested did not respond alike.
  • The setting. One study used a greenhouse with sunlight and supplementary LEDs. The other used a growth chamber with CO₂ at 1,200 ppm. A tent usually has neither sunlight nor added CO₂.
  • The medium and pot. Both used peat-based mixes in 2.5- or 6.7-litre pots. The tutorial uses coco in an 11-litre fabric pot. How quickly a flush strips nutrients depends on the medium and its volume, and coco was not tested.
  • The outcome. The trials measured yield, cannabinoids and nutrient concentrations. The main reasons small growers flush, taste, smoothness and burn, were not measured at all.
  • The economics. The calculation of $39 lost per $0.13 saved is about selling flower by weight. A home grower with one or a few plants may weigh a small loss of mass very differently against a finish they prefer.

None of this cancels the findings. The physiology, rinsed medium, deprived roots and leaves giving up their nutrients, applies to any cannabis plant. But the size of the effects, and whether there is any benefit to set against them, cannot simply be copied from hemp in a greenhouse to a THC cultivar in a tent.

What this means for a 1×1 tent

CannaTinker's position is that flushing stays a recommended small-grow practice, done briefly and with moderation. The research does not show that it improves taste, and it does not rule that out either. What it does show clearly is that the cost rises with length. That points towards a short, sensible flush rather than abandoning the practice or extending it.

  • Keep it short. The tutorial's 7–10 days sits within the range where losses were smallest. One week cost nothing measurable in 'Southern OG' reproductive mass, though 'THM Jack' lost about 0.01 kg/m² of flower per week from the start, and in 'Southern OG' the bigger losses came beyond two weeks. There is no evidence that starting earlier or stretching the flush brings anything extra.
  • Leach to a sensible target, not to zero. The tutorial aims for a runoff EC below about 0.6 mS/cm, close to the Utah target of 0.5 and gentler than Clemson's 0.2, which needed about six pot volumes of water. Check the EC of your own tap water. Water that is itself salty limits how far the medium can be rinsed.
  • Feed to demand beforehand. The less excess salt accumulates during flower, the less there is to rinse out, and the less dramatic the flush needs to be. Comparing the EC of feed and runoff through bloom is the simplest guide.
  • Read the leaves. Older fan leaves yellowing is expected: that is remobilisation. If upper leaves pale quickly, or yellowing races ahead in the first few days, do not extend the flush.
  • Consider the gentler variant. In one small trial, withholding only nitrogen did not reduce yield, while plain water did (Hershkowitz, 2024). It is a single study, but the only one that tested a middle path.
  • Taste is still your call. If you want to know whether flushing changes your flower, the honest test is a blind comparison of flushed and unflushed plants of the same cultivar, grown and cured the same way. That is exactly the experiment the research has not yet done.

What we know, and what we don't

We know, from two independent groups, that a thorough flush with plain water costs some flower mass, and that the loss grows with the length of the flush. We know that THC and CBD concentrations change little, with CBD sometimes rising slightly as flower mass falls. We know that flushed plants pull nutrients out of their older leaves, which is why fan leaves yellow. We know that flushing cannot wash anything out of the buds themselves.

We do not know whether flushing changes taste, smoothness or how the flower burns, because no controlled study has measured them. We do not know why the two groups disagree on flower nutrients. And we do not know how THC-dominant cultivars, grown in coco under a small LED in a tent without added CO₂, respond to a 7–10-day flush, because nobody has tested that combination. Earlier studies that found no effect on yield may simply not have deprived their plants.

That leaves room for practice. Flushing began, most likely, as a sensible fix for overfed plants, and it may or may not do what its fans say. Kept short and moderate, it costs little in a small grow. Stretched over weeks, the trials suggest it costs real flower for benefits nobody has yet measured.

Sources

Alden, M.J. (2025) Horticultural strategies to improve yield and efficiency for controlled-environment cannabis production. PhD dissertation. Clemson University. Available at: https://open.clemson.edu/all_dissertations/4161.

Alden, M.J. and Faust, J.E. (2025) 'Exploring the legacy practice of flushing in controlled-environment production of high-CBD cannabis (Cannabis sativa)', HortScience, 60(10), pp. 1818–1825. doi:10.21273/HORTSCI18752-25.

Hershkowitz, J.A. (2024) Nutrient management of cannabis in controlled environments. MSc thesis. Utah State University. Available at: https://digitalcommons.usu.edu/etd2023/178.