What's the strongest red light therapy you can use at home?

Irradiance is power per unit area at your skin, usually quoted in mW/cm². Dose is irradiance multiplied by time, usually quoted in J/cm². Most published protocols land in a broad band of a few joules to a few tens of joules per square centimetre.

This is why marketing wattage is close to meaningless. A 3,000 W panel spread over a wide area at a metre's distance can deliver less to your skin than a smaller, closer, better-focused device.

The trap in the other direction: a very high-irradiance device makes short sessions possible but can also make it easy to overshoot, since the response plateaus. Higher output should shorten sessions, not extend them.

Full-body beds sit at the top, because coverage does the heavy lifting. Even at moderate irradiance, treating most of your skin at once delivers far more total light than sequential spot treatment.

Large panels come next. A tall panel can cover a torso or a pair of legs per position, at high irradiance if it's well built — but you're repositioning, and each reposition costs time.

Wearables and masks are moderate by design, trading peak output for contact and convenience. Contact delivery is efficient, so their effective dose is often better than their specs suggest.

Handhelds are the lowest total dose per session, which is exactly right when you only need one joint or scar treated.

Ask for irradiance at a stated distance — 15 cm and 30 cm are common reference points — and compare at the same distance. Surface-of-LED figures aren't comparable to anything.

Compare the treatment area at that distance too. A high number over a small window is not the same product as a slightly lower number across a whole body.

Check the wavelength tolerance. Binned LEDs hold their stated wavelength; loose tolerances mean part of your output is landing outside the useful range.

Then divide by time you'll realistically spend. The strongest device you never get into is weaker than the moderate one you use five times a week.