Short answer: it is not the colour of a lamp that decides whether it disturbs your body clock, but how much light lands in the range the circadian receptor in your eye is sensitive to: around 480 nanometres. The measure for that is called melanopic EDI. Two lamps that look equally bright and equally warm can differ by a factor of ten on it.
Why “blue light” is too rough a term
Besides rods and cones, your retina contains a third kind of light sensitive cell: the intrinsically photosensitive retinal ganglion cell, ipRGC for short. This cell contains the pigment melanopsin and is not there to form images, but to tell your brain what time it is. Melanopsin is most sensitive around 480 nm, cyan blue.
Hence the rule of thumb to avoid blue light in the evening. Broadly speaking that rule holds, but it is too rough for three reasons.
- “Blue” in the usual sense runs from about 450 to 495 nm. Circadian sensitivity peaks in a narrower band within that, and according to current science light at 460 to 500 nm counts more than light at 430 nm.
- Your ordinary cones have a say too, especially in the first minutes of an exposure.
- Amount and duration count. A weak lamp with a bit of blue in it is not the same as a bright screen with the same amount of blue.
So judging a lamp on colour alone is not enough. You need a measure that combines wavelength and amount.
What the research shows
Sensitivity shifts during the exposure
The spectral sensitivity of human circadian phase resetting and melatonin suppression to light changes dynamically with light duration, St Hilaire et al., PNAS, 2022. Ninety-nine healthy young participants (34 women, 65 men, average age 23) received 6.5 hours of continuous monochromatic light during their biological night. For each wavelength, the researchers measured how much melatonin was suppressed and how far the body clock shifted.
The result is more subtle than the rule of thumb. In the first quarter of the exposure, peak sensitivity for melatonin suppression was at 441 nm, with a second peak around 550 nm. That points to a strong contribution from the ordinary cones: both the blue sensitive S cones and the L and M cones that see green and red. In the last quarter that contribution had faded and the peak was at 485 nm, almost exactly where melanopsin sits.
In practice: with short exposures your whole visual system takes part and green and yellow light count too. Only with longer exposure does melanopsin take over. So “just remove the blue” is a simplification, and if you get a short, bright burst of light, that measure alone does not cover you.
Limits. Young, healthy participants; monochromatic light in a lab is not the same as a lamp in your living room; and the exposure lasted 6.5 hours, longer than an average evening.
At equal brightness and colour, melanopic content makes the difference
Melanopic irradiance defines the impact of evening display light on sleep latency, melatonin and alertness, Schöllhorn et al., Communications Biology, 2023. Seventy-two healthy men were exposed to light at four brightness levels four hours before their usual bedtime, matching dimmed light, a smartphone, a tablet and a computer screen. At each level there was a version with low and one with high melanopic irradiance.
The clever part of this design: brightness and perceived colour were kept the same. The only thing that differed was how much of the light fell in the melanopic range. In other words, two lamps that look identical to your eye.
With the low melanopic version, participants fell asleep faster, melatonin was suppressed less in the evening, melatonin levels were higher the next morning, melatonin release started earlier in the evening, and participants were less alert. The effects also depended on the dose: the higher the melanopic irradiance, the longer it took to fall asleep and the further melatonin onset shifted.
For subjective alertness, what people reported themselves, there was no such dose relationship.
Limits. Men only, which limits how far this applies to women. Healthy adults in a controlled setup. And it was about screen light for four hours, not a night light for ten minutes.
The measure that makes this measurable: melanopic EDI
That research leads to a practical conclusion. If the effect depends on the melanopic content of the light, you need to be able to measure it, and there is an international standard for that: CIE S 026, which defines the so called α-opic quantities.
The most important of these is melanopic EDI, short for melanopic equivalent daylight illuminance, expressed in lux. It tells you: how much daylight would you need to give the melanopsin system the same signal as this lamp? Then there is melanopic DER, the ratio between melanopic EDI and ordinary illuminance. These numbers can be compared between lamps, regardless of their colour or stated wattage.
| Measure | What it tells you | Why you need it |
|---|---|---|
| Illuminance (lux) | How bright the light is to your eye | Says nothing about the circadian effect |
| Colour temperature (K) | How warm or cool the light looks | Two lamps with the same K can differ a lot in melanopic content |
| Melanopic EDI (lux) | How much signal the melanopsin system gets | This is the measure the research uses |
| Melanopic DER | Melanopic EDI divided by illuminance | Makes lamps of different brightness comparable |
| Emission below 480 nm | What share of the light falls in the short wavelength range | Can be checked directly on a spectral curve |
And here is the problem for anyone choosing a lamp: sellers of night lighting almost never publish these numbers. You get wattage, colour temperature and sometimes lumen, exactly the values that say nothing about the question you are asking. But since most people are not scientists and simply want good anti blue light, you can also look at the light curve images shown on our anti blue light lamp product pages and pick a lamp that suits your situation.
What this tells you about evening lighting
- Ask for the melanopic EDI at the distance you use it, not for lumen or kelvin. If you cannot get it, the lamp has not been assessed on this point.
- Look at the spectral curve, not the colour of the lamp. A lamp can look warm and still have a peak below 500 nm.
- Brightness counts. Dimming lowers the melanopic signal proportionally. A lamp with an unfavourable spectrum that burns very dimly can give less of a signal than a favourable lamp at full strength.
- Duration counts. The research works with hours of exposure. A few minutes of light is a different situation from a whole evening.
- Placement counts too. Light coming from below or from the side reaches the lower part of your retina differently than a screen right in front of your face.
Finally
- The research was not done on everyone. The two studies above used young, healthy adults; the second only men. These results say less about children, older people, shift workers and people with eye conditions.
- Lab light is not living room light. Monochromatic light in a controlled setup behaves differently from a lamp among other light sources, with reflections and changing viewing directions.
Questions that build on this
- Which night light won’t disturb my sleep? The four properties that count, in order of importance
- Is red light in the evening better than dimmed white light? Why the advantage of red is real but limited
- Lighting for night feeds and night shifts Why the advice for these two situations is opposite
- Blue light free lighting for a child’s bedroom Children react more strongly, and dimming matters more than colour
- Reading without disturbing your body clock Paper, e-reader or screen, and what night mode does and does not do
Sources
- St Hilaire et al., The spectral sensitivity of human circadian phase resetting and melatonin suppression to light changes dynamically with light duration, PNAS, 2022
- Schöllhorn et al., Melanopic irradiance defines the impact of evening display light on sleep latency, melatonin and alertness, Communications Biology 6:228, 2023
- CIE S 026: the international standard for α-opic quantities, including melanopic EDI
Last reviewed: 21 August 2026. This page is updated when new research appears that affects the conclusions.
