Nutrient film technique: how it works and how to build one
A thin, continuous film of solution runs past the roots and back to a reservoir. Very water-efficient, very compact, and less forgiving than it looks.

NFT is what most commercial lettuce is grown in. A shallow stream of solution, only a few millimetres deep, runs down a gently sloped channel. Roots sit in the channel with their tips in the film and the rest of the root mass in humid air above it. Almost everything that goes wrong with an NFT build comes from one of three numbers being wrong, and all three are set when you build the frame rather than when you plant.
A film, not a bath
This is the part that people get wrong, and getting it wrong turns an NFT channel into a long thin flooded trough that drowns roots slowly.
The solution is supposed to be a film a few millimetres deep. Only the lower part of the root mass sits in it. The rest of the roots hang in the humid air above, and that is where most of the oxygen comes from. Roots in moving water get fed. Roots in air get to breathe. Doing both at once, continuously, is the entire idea.
Oregon State Extension notes that channels sometimes have textured bottoms or a woven mat laid along them, which is there to spread the film across the full width instead of letting it find one path down the middle. It is a detail worth copying in a home build, where a length of capillary matting costs very little and does the same job.
The first number: slope
Too flat and the solution pools around the roots, which suffocates them. Too steep and the film runs past too fast and too thin to feed anything. Oregon State’s guidance is a minimum of 2 percent, and it adds that you should check the slope periodically rather than setting it once, because frames move.
Two percent is a ratio, which is useless when you are holding a spirit level and a length of pipe. Here it is in the units you are actually working in.
| Channel | Minimum | Safer |
|---|---|---|
| 3 ft | 0.7 in | 0.9 in |
| 4 ft | 1.0 in | 1.2 in |
| 6 ft | 1.4 in | 1.8 in |
| 8 ft | 1.9 in | 2.4 in |
Total drop from the inlet end to the outlet end. Minimum is the 2 percent Oregon State gives. Safer is one in forty, which is what commercial guidance tends to use in practice.
Why build steeper than the stated minimum? Because the textbook ideal is actually much flatter, around one in a hundred, and the only reason practical guidance recommends one in thirty to one in forty is that nobody can build a base true enough for the flatter figure. A slight dip in a home-made frame becomes a puddle, and a puddle is a dead patch of roots. The extra slope buys you tolerance for a frame that is not perfect, which yours will not be.
The second number: flow rate
The figure you will see quoted everywhere is one litre per minute per channel, with about two litres as the upper limit and roughly half that at planting. That is sound, and it is worth knowing that it is a commercial figure, written for a long gully carrying dozens of plants.
A home channel is a different animal. Published guidance aimed at smaller systems runs considerably lower, in the region of a few hundred millilitres per minute per channel for lettuce. If you are running six plants in a four foot length of pipe, start at the bottom of the range rather than the top.
The practical test does not need a flow meter. Look at the channel. You want a sheet of water moving quietly along the floor of it. If you can see it rippling, splashing at the inlet, or running in a deep central stream, there is too much. If there are dry patches between the plants, there is too little. A tap or a simple valve on the feed line is worth more here than a precisely sized pump, because it lets you tune by eye.
The third number: channel length
Solution changes as it travels. By the time it reaches the far end of a long channel the plants upstream have taken things out of it, and the published observation is that growth falls off noticeably beyond about twelve metres, with ten to fifteen metres given as the practical ceiling.
What runs out first is worth knowing, because it is not what people assume. On fast growing crops the reported problem along the length of a gully is nitrogen depletion rather than oxygen. The commercial fix is to feed the channel again halfway down rather than to push more solution through from one end.
At home this is almost never your limit. A four or six foot channel is nowhere near the distance at which any of this matters, and if you are scaling up, the right move is more channels fed in parallel rather than longer ones fed in series. That is also how NFT is designed to grow, and why commercial installations are banks of short gullies rather than a few long ones.
The fourth number, and the only one you set at planting
Slope, flow and length are all fixed when you build the frame. Spacing is the one you choose again every time you plant, and it is the one people get wrong in the opposite direction, by filling every hole the channel happens to have.
The range growers have settled on is six to eight inches between plants along the channel for loose leaf and butterhead lettuce, eight to ten for a full romaine, and eight to twelve inches between channels so the canopies do not close over one another. Herbs sit at the bottom of that range, although a basil plant will use all eight inches if you give them to it.
| Length | At 6 in | At 8 in |
|---|---|---|
| 3 ft | 6 | 4 |
| 4 ft | 8 | 6 |
| 6 ft | 12 | 9 |
| 8 ft | 16 | 12 |
Plant positions in one channel. Arithmetic from the spacing, not a yield figure.
Those are positions rather than plants to fill on day one. A channel planted all at once gives you every head in the same week, which is the problem the article on succession planting exists to solve, and a channel is the easiest place on this site to run a rotation because the holes are already spaced and numbered by the build.
Crowding costs more here than it does in a bucket. In a bucket, close planting competes for light and oxygen. In a channel it also narrows the gap the film has to flow through, which is the root mat problem below arriving early and all at once.
Treat these as the weakest numbers on this page. The slope, flow and length figures above come from extension and commercial literature. The spacing range is what growers and channel manufacturers have converged on, which is worth knowing and is not the same kind of evidence.
The problem nobody warns you about
A few millimetres of water spread thinly across a plastic channel has almost no thermal mass. It takes the temperature of the room, fast, and it does it along the whole length of the run where the plants are.
Oregon State flags this directly, that shallow NFT channels are sensitive to ambient temperature swings and that this causes nutrient imbalances and plant stress. This is the quiet disadvantage of the design and it rarely appears in beginner guides at all, because it does not announce itself the way a pump failure does.
Their three suggested mitigations are worth repeating exactly, because two of them are counter-intuitive.
- Increase the slope, so solution spends less time in the channel being warmed or chilled by the room.
- Shorten the channels, for the same reason.
- Use a larger stock tank, so the bulk of the solution sits in a reservoir with real thermal mass rather than spread out in the open.
That last one is the easiest and it is free if you simply buy a bigger tub at the start. It is the same argument made in the article on reservoir changes, for a different reason, and it is why a generous reservoir is the best money in this build.
What it does well
- Uses less water and nutrient than any other recirculating design
- Very high plant density in a small footprint
- Easy to inspect and harvest, because plants lift straight out
- Scales neatly by adding parallel channels
Where it struggles
- A pump failure dries the roots within a few hours
- Root mass can block the channel and cause overflows
- Poorly suited to heavy fruiting crops
- Needs a level, stable mounting structure
- Thin film takes the room’s temperature quickly
The failure mode to plan for
Oregon State puts the same point more bluntly, that given the system’s reliance on a thin film, backups for power and pumps are crucial, and that a loss of either leads to immediate wilting. Immediate is the operative word. There is no grace period built into this design the way there is in a bucket. The only design on this site with a shorter one is aeroponics, where the roots have nothing on them at all.
Three things are worth doing before you trust it unattended. Run the pump off a socket you do not share with anything you might switch off at the wall. Check it visibly every day, which takes two seconds because a running channel looks different from a stopped one. And if the system is ever going to be left for more than a day, put a second pump in the tank on the same feed, so a failed impeller is an inconvenience rather than a total loss.
The root mat, and when it turns on you
Roots in an NFT channel do not spread out into a medium, because there is none. They form a dense mat along the floor of the channel, and that mat is doing useful work for most of the crop, holding the film and spreading it across the width.
Late in a crop it becomes a dam. The mat gets thick enough to hold water behind it, the film backs up into a pond, the roots upstream of the blockage sit in standing water, and you get the one thing the design exists to avoid. In a bad case the channel overflows at a joint.
This is why channels offer limited rooting space and suit short lived, small rooted crops, and it is the real reason tomatoes and peppers are a poor fit rather than anything to do with their nutrient demands. It is also an argument for harvesting on time rather than letting a crop stand, which the article on when to harvest makes for entirely separate reasons.
Lift the far end of each channel occasionally and look down it. A clear floor means the film is moving. Standing water above a wall of root means it is time to pull the crop.
What to grow
The crops Oregon State lists for NFT are lettuce, herbs including basil, cilantro, mint and parsley, spinach, arugula, kale, Swiss chard, mustard greens, and certain varieties of strawberry which can be run as a perennial crop in either horizontal or vertical systems.
That list is the design telling you what it is for. Everything on it grows fast, grows shallow, and comes out before the root mat becomes a problem. Basil and lettuce are the two that do best of all. Strawberries are the interesting exception, because they are a long crop, and the reason they work anyway is that their roots stay comparatively modest and their fruit hangs clear over the channel edge.
Avoid tomatoes and peppers. They will grow for a while and then dam the channel, and a large fruiting plant in a shallow pipe is also top heavy in a way the structure will not enjoy.
Does it work in aquaponics?
Yes, and it needs two pieces of equipment a media bed does not. The reason is worth understanding before you buy anything.
In aquaponics the fish supply the nutrients, but not in a form a plant can take up. Bacteria do that conversion and they need surface area to live on. Arizona Cooperative Extension is direct about what follows: solids removal and biofiltration are essential, and the water goes from the fish tank through both of them before it reaches the plants. In a media bed the gravel or clay provides that surface area as a side effect of being there. An NFT channel provides almost none, so the biofilter has to be a separate vessel you buy, plumb and maintain.
Solids are the other half, and they matter more here than in any other design. Uneaten feed and fish waste produce suspended and settleable solids, and a channel carrying a film a few millimetres deep is the worst place in a system for them to come out of suspension. They coat roots, they build up at the low end, and they turn the film into exactly the flooded trough this page opened by warning about.
So an aquaponic NFT runs fish tank, then solids filter, then biofilter, then channels, in that order. It works, and it keeps the water efficiency that makes NFT attractive. It is simply not the aquaponic system to learn on, and for a first build a media bed asks less of you because the medium is doing two jobs at once.
Build one
Our $50 PVC pipe plan is a six to eight plant NFT build, including the slope setup and the plain-water test run we recommend before you plant anything. Run it on plain water for a full day first and watch where the film actually goes, because that is when a low spot or a joint that weeps is cheap to fix.
Not sure which system to build?
Compare all six side by side on cost, difficulty, and what they grow best.
Where this comes from
The minimum 2 percent slope, the advice to re-check it, the warning that pump or power failure causes immediate wilting, the temperature sensitivity of shallow channels and its three mitigations, the note on textured channel bottoms, and the crop list are from Oregon State University Extension’s peer reviewed publication EM 9457, Hydro hints: nutrient film technique, published May 2025. The flow rates of about one litre per minute per gully with two litres as an upper limit, the one in thirty to one in forty practical slope against a one in a hundred ideal, and the channel length ceiling with nitrogen depletion along the gully are long standing figures from the commercial NFT literature. The drop figures in the table are arithmetic from the percentages rather than measurements. Where home scale guidance differs from commercial guidance, as it does on flow rate, that is said rather than averaged.
The aquaponics section is from University of Arizona Cooperative Extension’s peer reviewed publication az2215, Aquaponics Basics: How Fish, Plants, and Microbes Work Together, which states that solids removal and biofiltration are essential, that water flows from the fish tank through both before reaching the plant growing area, that uneaten feed and fish waste produce suspended and settleable solids, and that a media bed uses growing media which also acts as biofiltration. The plant spacing range is not from either of those and is labelled above for what it is, the figure growers and channel makers have converged on. The plants per channel table is arithmetic from that range.
Answer five questions about your budget, space, experience and what you want to grow, and our system comparison tool ranks all six against your answers, with the reasoning shown, and an honest “nothing fits” when nothing does.
