Fluorophore brightness comparison
Rank fluorophores by what your setup will actually detect, not by ε × Φ alone.
| Fluorophore | Em | ε | Φ | ε × Φ | Excited | Collected | In this setup | Bleach t½ |
|---|---|---|---|---|---|---|---|---|
| mStayGold | 510 | 164k | 0.83 | 136 | 60% | 80% | 100% | — |
| mNeonGreen | 517 | 116k | 0.80 | 92.8 | 62% | 75% | 66% | 158 s |
| Fluorescein (FITC) | 517 | 88k | 0.92 | 81 | 77% | 68% | 66% | — |
| BODIPY FL | 511 | 83k | 0.99 | 82.2 | 55% | 78% | 55% | — |
| Alexa Fluor 488 | 520 | 73k | 0.92 | 67.2 | 75% | 69% | 54% | — |
| mEmerald | 509 | 58k | 0.68 | 39.1 | 100% | 67% | 41% | 101 s |
| EGFP | 507 | 56k | 0.60 | 33.5 | 100% | 67% | 35% | 174 s |
| mCitrine | 529 | 94k | 0.74 | 69.6 | 36% | 61% | 23% | 33 s |
| mVenus | 527 | 104k | 0.64 | 66.6 | 36% | 62% | 23% | 50 s |
| EYFP | 527 | 67k | 0.67 | 44.9 | 38% | 61% | 16% | 60 s |
| mOrange2 | 565 | 58k | 0.60 | 34.8 | 20% | 5% | 1% | 228 s |
| mCerulean3 | 475 | 40k | 0.87 | 34.8 | 3% | 35% | 0% | — |
| mKO2 | 565 | 64k | 0.62 | 39.6 | 17% | 5% | 0% | 5 s |
| ECFP | 477 | 33k | 0.40 | 13 | 5% | 38% | 0% | 85 s |
| Janelia Fluor JF549-HaloTag conjugate | 576 | 83k | 0.81 | 67.4 | 7% | 3% | 0% | — |
| Cy3 | 566 | 136k | 0.15 | 20.4 | 15% | 4% | 0% | — |
| Alexa Fluor 546 | 572 | 112k | 0.79 | 88.5 | 8% | 2% | 0% | — |
| mTurquoise2 | 474 | 30k | 0.93 | 27.9 | 1% | 36% | 0% | 90 s |
ε × Φ is the fluorophore’s own brightness, in the usual units of 1000, and is the number papers tabulate. In this setup multiplies it by how well your line excites the molecule and how much of its emission your filter passes, relative to the best row on show. Neither accounts for maturation, folding, expression level or how well the fusion tolerates the tag — in a real cell those routinely matter more than either column.
Bleach t½ is seconds to half the initial emission, as published. It is deliberately not sorted on and the rows are not ranked by it: these figures are not comparable with one another. Each comes from a different paper at a different illumination intensity, in a different medium, on a different objective, and the same protein has been published with values an order of magnitude apart. Treat a small number as a flag to test the fluorophore yourself, not as evidence it is worse than the row above. Dyes carry none here at all.
ε × Φ ÷ 1000; in this setup = ε × Φ × EX(λ) × ∫EM(λ)T(λ)dλ ÷ ∫EM(λ)dλ- Fluorophore spectra, extinction coefficients and quantum yields — FPbase, Nature Methods, 2019
- Side-by-side comparison of fluorescent protein brightness and photostability in cells — Nature Methods, 2016
- Choosing a fluorescent protein, and how photostability is measured — Nature Methods, 2005
Calculated in your browser. Nothing you entered was sent anywhere.
When to use this
Use this when deciding which fluorophore to put in a construct or on an antibody, and you want to know what your own microscope will detect rather than what a table says. Molecular brightness is the fluorophore’s own property; practical brightness folds in how well your line excites it and how much of its emission your filter passes. Use the spectra viewer instead if the question is about colour separation rather than signal.
Worked example
A widefield green channel: a 488 nm line and a 525/50 emission filter.
- Laser
- 488 nm
- Emission filter
- 525/50
- Rank by
- In this setup
Result
mEmerald is the only one of the four that this line excites fully, at 100%, and it still finishes last: being perfectly matched to the laser does not compensate for having a third of the extinction coefficient.
What people get wrong
- Choosing on ε × Φ alone. Move the same comparison to a 405 nm line and a 450/50 filter and mCerulean3 is the brighter molecule while mTagBFP2 delivers 3.9 times the signal, purely because the line excites it to 96% rather than 58%.
- Ranking fluorophores by the photobleaching half-lives shown here. Those figures are not comparable with one another: each comes from a different paper at a different illumination intensity, in a different medium, and the same protein has been published with values an order of magnitude apart. Read a small number as a prompt to test it yourself.
- Forgetting everything this does not model. Maturation time, folding efficiency, expression level, and whether the fusion tolerates the tag at all routinely matter more in a live cell than any column here — a bright protein that matures slowly is dark for the first hours of a timelapse.
- Assuming a dye conjugate behaves like the free dye. These figures are for the free fluorophore; on a densely labelled antibody, self-quenching can cost most of the brightness, and it is worst for exactly the narrow-Stokes-shift dyes that look best on paper.
Questions
+What are the units of brightness?
ε × Φ divided by 1000, which is the convention every fluorescent protein paper uses — EGFP is 33.5 on that scale. It has no physical meaning on its own and exists only to be compared with other entries in the same column.
+Why is a bright far-red dye shown at zero in my green setup?
Because a 488 nm line does not excite it. The practical column is the molecular brightness multiplied by how much of the molecule the illumination reaches, so a fluorophore the laser cannot touch scores near zero however bright it is in principle.
+Does this account for the camera?
No. Detector quantum efficiency is not modelled, which is fine across the visible range and increasingly wrong past about 700 nm, where a silicon sensor falls off steeply. A near-infrared dye will look better here than it will on your camera.
+Should I always pick the brightest one?
No. Photostability decides long timelapses, monomeric behaviour decides whether a fusion works at all, and maturation decides what you see in the first hours. Brightness matters most when signal is genuinely the limit — which is worth confirming before optimising for it.
Related tools
- Fluorescence spectra viewer — Overlay excitation and emission spectra and see which pairs will separate.
- Filter set and channel checker — Check a multi-colour panel against your filters before you stain anything.
- FRET pair calculator — Förster radius from real spectra, and the artefacts that will spoil the measurement.
Science last reviewed .