Patients bring me these.

That is genuinely how this started. I treat backs, hips and knees, and over the last two years enough people have arrived holding a red light belt and asking whether it is any good that I finally sat down and went through them properly.

I am not paid to like any particular one, and I am going to tell you plainly which parts I could not verify.

The short version: they are not in the same category as each other, and the difference does not show up until about week three, which is exactly when it is too late to send anything back.

The Mistake Almost Everybody Makes First

They compare on price and star rating. That is the whole method most people use, and I understand why, because there is very little else to go on.

• They all feel identical on night one. Every one of these arrives, switches on, glows red and feels warm. There is nothing to learn from the first night.
• The listings are near-identical. Same wavelengths printed, same claims, frequently the same photographs with a different logo on them.
• The reviews are front-loaded. Most are written in the first fortnight by people who cannot yet know anything.
• The return window usually closes before the answer arrives. Fourteen to thirty days on the cheap ones, against a mechanism that takes considerably longer than that.

Which means the comparison has to happen before you buy

You cannot test your way to the answer afterwards, because the test takes longer than the window you are given to run it. So you have to compare on things you can actually check from a listing, and there are fewer of those than you would hope.

What follows is what I now check, in the order I check it. It is not a scoring system and I am not ranking every belt on the market, because nobody has tested every belt on the market.

It is five questions. Two of them will be answered on any honest listing. Three of them will usually not be answered at all, and that silence is itself the most useful information you will get.

One. The Split, Not The Wavelengths

Every belt on the market prints 660nm and 850nm. That number settles nothing, because they all say it. The question is how many of each.

Why the ratio is the whole thing

660nm red is the one working at the skin layer. 850nm infrared travels further, toward muscle. Most of these are built infrared dominant, because most of them are designed for muscle and joint pain. That is a perfectly good product. It is simply a different product, and nobody selling it will tell you which one you are holding.

What happened when I asked

Two. Output, And What I Could Not Verify

I asked several sellers directly how many chips of each. The reply I usually got was some version of “professional grade 660nm and 850nm LEDs”, which is not an answer to the question I asked. One did not reply at all. The one that answered runs two red chips to every one infrared: 315 chips, 210 red and 105 infrared, and prints the split on the box.

Now the part I want to be straight about

I could not build an honest comparison on power output. Almost nobody publishes real irradiance figures, and where they do there is no independent way to check them.

That includes the one I would buy. I am not going to quote you a number I cannot stand behind, and I would treat any brand quoting a big power figure without a test report exactly the same way.

So I dropped output as a criterion entirely. It is the criterion that sounds most scientific and it is the one with the least available evidence behind it, which is an uncomfortable thing to discover halfway through.

I would look at photographs like that one the way I look at any before and after, which is to say checking the lighting and the posture first. The only version that counts is your own waistband at week eight.

That small box is what I would look at hardest. It is what had failed on two of the belts patients brought me, and it is the cheapest component on any of them to make badly.

From the clinic

I am not going to tell you a device will change your body. What I can tell you is what I have watched, across a lot of people, over several years, and which of these were still working at the point where they were supposed to start working.

A licensed physical therapistMusculoskeletal practice

60 nights to decide. If it has not changed anything by then, email us and we refund you. The 1-year warranty is included either way.

Try ThermaLight360 →

Three. The Warranty, Because It Is A Prediction

This is the most useful thing on this page and it took me far too long to see it. A company’s warranty is its own private estimate of when its product will fail. The cheap ones run thirty days. That is not stinginess, it is arithmetic. They know.

The one I would buy carries a year, included, with a sixty day money-back window on top of it. You are not reading a warranty. You are reading a confession. And it is printed on the box, for free, before you have spent anything. I have started asking patients to photograph the warranty line before they order, because it is the one piece of information a seller cannot spin and cannot quietly omit once it is in writing. If a listing does not state a term at all, assume the shortest one the platform allows, because that is what it will turn out to be when you need it. And notice which way the incentive runs: a company expecting to still be selling this in three years has every reason to stand behind it for one, while a company expecting to have moved on has every reason not to.

Four. Whether It Survives Long Enough To Be Judged

Here is the part that reframes almost every disappointed person I meet. The layer this works on sits roughly an inch and a half down, and it remodels on a schedule. Six to eight weeks before anything is fair to assess.

The belts patients have brought me failed in week two, week three, week four. One arrived with a control unit that had already died. Those were mathematically incapable of working. Not underpowered, not badly designed. They quit before the clock they were being judged against had started.

Which means the person in front of me has usually drawn the wrong conclusion. She thinks she failed. She never got far enough in for it to be a fair test, and she has been carrying that for months. I want to put a number on how common this is, because it surprised me. Of the belts patients have physically handed me, not one was still functioning at the eight week mark. Not a single one. That is a small sample and I would not publish it as research, but it is also not nothing, and it changed how I answer the question in the room.

• Five. Length, and cost per session. Cheap belts run short. A belt you have to hold in place with one hand is not a routine anybody keeps, and I have watched that single detail end more attempts than any other.
• 50.8 inches fastens on most people. Measure yourself before you buy anything, and if a listing will not state a length, that is your answer.
• Three or four cheap belts at around forty dollars comes to more than one good one and produces nothing, which makes the cost per useful session impossible to calculate.
• $139.95 once, used most nights across a year, is pennies a session. The cheap belt is not the cheap option. It is the one that takes the money more slowly.

Who I Would Actually Say This Is For

Not everybody, and I would rather say so than take a sale I do not believe in. Three groups arrive in my room with this and they are quite different from one another.

People who already own one and think they failed

This is the largest group and the one I most want to reach. Go and find two things: your belt’s warranty length, and whether the seller will tell you the red to infrared split. Those two answers will tell you whether you were ever running a fair test.

People about to buy their first one

Do not sort by price. Sort by whether it will still be running in week eight, because week eight is the first honest checkpoint and everything before it is guesswork.

People who want it for pain rather than skin

Then an infrared dominant belt may genuinely suit you better, and I would say so. Different job, different tool. I am not going to pretend one device is the answer to two different questions.

And the group I would tell to wait: anybody who needs an answer inside a month. The mechanism does not move at that speed and no belt on the market changes that. I would rather lose the sale than have somebody buy on Tuesday for an event on Saturday, because that person is not going to get a result and they are going to conclude the whole category is a con. That conclusion is expensive for them and it is unfair on the one product in the drawer that might actually have worked given a fair run at it.

The Side By Side, Written Out

Wavelengths printed: 660nm and 850nm on both. Everyone says this and it settles nothing. The split, red to infrared: not published on the cheap ones, and sellers would not answer when asked. Published on the one I would buy, at 210 red to 105 infrared, two to one, red dominant.

Total chips: not stated on the cheap ones. 315 on the other. Irradiance: not published anywhere, by anyone, including the one I would buy, which is why I did not score it.

Warranty: thirty days against one year included. Money-back window: fourteen to thirty days, shorter than a result takes, against sixty nights. Length: often too short to fasten and rarely listed, against 50.8 inches, listed.

Survived to week eight: the ones brought to me did not. The other one did. Price: about forty dollars several times over, against $139.95 once.

What I tell patients

If you tried one of these and it did nothing, please consider that you may not have failed at all. You may simply have been holding something that stopped before the test began.

A licensed physical therapistMusculoskeletal practice

60 nights to decide. If it has not changed anything by then, email us and we refund you. The 1-year warranty is included either way.

Try ThermaLight360 →

The Conclusion People Draw About Themselves

Nearly everybody who brings me a dead belt says a version of the same sentence, which is that they never stick at anything.

They stuck at it for a month. The belt did not last a month.

It was not their consistency that failed, it was the hardware. I find that genuinely irritating, because the belief that they are the problem is considerably harder to undo than the purchase was.

Twenty minutes a night is not a discipline problem. It is short enough that being undisciplined does not really matter, which is rather the point of a device you strap on and forget.

These are the sort of notes I get afterwards, and they have a shape in common. Nobody reports a measurement to me. They report something they started or stopped doing.

★★★★★

Angela, 54

I had written it off entirely. It turned out my old one had a thirty day warranty and I had been blaming myself for two years about a thing that was never going to work.

★★★★★

Marie, 56

The warranty question was the one that did it for me. I looked at the box of the one in my drawer and it said thirty days and I understood immediately.

★★★★★

Denise, 51

Two died on me before this. This is the first one still working at the point where it was supposed to start working.

Not one of them gave me a number, which after several years of asking I have started to find more convincing rather than less.

What The Extra Money Is Actually Buying

• The ratio, printed.210 red chips to 105 infrared, two to one, red dominant, across 315 chips. Most belts are built the other way round because most are built for pain, and almost none will tell you which you are holding.
• A control unit that survives.The part that failed on the belts brought to me. It is the cheapest component to make badly and the one you cannot inspect from a listing.
• A year of warranty, included.Against thirty days. Read it as the company's own estimate of when it expects the thing to break, because that is what it is.
• Sixty nights to judge it.Longer than the mechanism takes, which is the only version of a guarantee that means anything on a product like this.

The Questions I Get Asked In The Room