How heavy is a neutrino?
Neutrinos are the lightest known particles with mass, at least a million times lighter than the electron. Their exact mass is still unknown, and three very different methods are closing in on it from different sides.
Calculator: the three neutrino masses →
Beginner
A particle that was supposed to weigh nothing
For decades physicists assumed that the neutrino had no mass at all, like the particle of light. That changed around 1998, when experiments found that neutrinos change their identity while they travel. There are three kinds, or flavours, of neutrino, and a neutrino born as one kind can later be caught as another. This effect, called neutrino oscillation, is only possible if neutrinos have mass. So today we know for certain that they are not massless. What we do not know is how heavy they are.
Knowing the differences, not the weights
Oscillation works a bit like two tuning forks that are slightly out of tune: you hear a slow beat, and the speed of the beat tells you how different the two pitches are, but not what the pitches themselves are. In the same way, oscillation experiments measure the differences between the masses of the three neutrino types very precisely, but not the masses themselves. From these differences we can at least say that one of the neutrinos must weigh a minimum amount, roughly one ten-millionth of the electron's mass.
Weighing with a scale made of electrons
The most direct way to weigh a neutrino uses beta decay, a kind of radioactivity. When an atom of tritium, a heavy form of hydrogen, decays, it releases an electron and an antineutrino that share a fixed amount of energy. Part of that energy is needed to create the neutrino's mass. If the neutrino has mass, the fastest electrons come out very slightly slower than they would otherwise. Experiments therefore measure the energies of electrons near the upper end of the spectrum with extreme care.
The German experiment KATRIN in Karlsruhe does exactly this. Its centrepiece is a vacuum tank as large as a house, 23 metres long, which filters electrons by their energy. In 2025 KATRIN reported that the neutrino mass is less than 0.45 electronvolts1. An electronvolt is a tiny unit of energy that physicists also use for mass; the electron weighs about 511,000 of them. So a neutrino is at least a million times lighter than an electron.
Weighing with the whole universe
A completely different method uses the universe itself. The Big Bang left behind a sea of relic neutrinos, about 340 in every cubic centimetre of space. Because there are so many, even a tiny mass changes how galaxies clumped together over billions of years. By mapping the oldest light in the sky and millions of galaxies, cosmologists find that all three neutrino masses together add up to less than about 0.06 to 0.12 electronvolts67. These limits are stronger than the laboratory result, but they depend on assumptions about how the universe works.
Which neutrino is the heaviest
Another open question is the order of the three masses, called the mass ordering. Either two light neutrinos sit below one heavier one, or one light neutrino sits below two heavier ones. The Chinese experiment JUNO, which began measuring in 2025, is designed to answer this by studying antineutrinos from nuclear power plants over several years.
Why it matters
Neutrino mass is the first clear sign that the Standard Model, the theory of particles, is incomplete, because the original theory gives neutrinos no mass. Knowing the actual value would help reveal where that mass comes from, and it would also tell cosmologists how much neutrinos shaped the structure of the universe.
Quiz
Test your knowledge
4 questions on this level · Print as worksheet
Intermediate
What oscillations tell us
Neutrinos are produced and detected as flavour states (electron, muon, tau), but they travel as three mass eigenstates with masses m₁, m₂ and m₃. Because these states move with slightly different phases, the flavour mixture changes along the way. The discovery of this effect by Super-Kamiokande in 1998 and SNO in 2001–2002 established that at least two neutrinos have mass. Oscillation experiments are sensitive only to mass-squared differences: today Δm²₂₁ ≈ 7.5 × 10⁻⁵ eV² and |Δm²₃₁| ≈ 2.5 × 10⁻³ eV²8. Taking square roots, at least one neutrino must weigh about 0.05 eV or more, and a second one at least about 0.009 eV.
Direct measurement with tritium
In beta decay the decay energy is shared between the electron and the antineutrino. Enrico Fermi already noted in 1934 that a neutrino mass would change the shape of the electron spectrum near its endpoint. Tritium is the ideal source: its endpoint energy is low, about 18.6 keV, its half-life is a convenient 12.3 years, and its atomic structure is simple. The method is purely kinematic and makes no assumption about whether neutrinos are their own antiparticles. It measures an effective mass, the effective electron neutrino mass mβ, an average of the three masses weighted by how much electron flavour each contains.
From Mainz and Troitsk to KATRIN
In the 1990s and 2000s two experiments in Mainz and Troitsk introduced the MAC-E filter, an electrostatic spectrometer that lets only electrons above a set energy pass. Mainz published a final limit of 2.3 eV in 20053 and Troitsk 2.05 eV in 20114, both at 95 % confidence. Their successor KATRIN, at the Karlsruhe Institute of Technology, scaled the idea up: a 70 m long beamline with a windowless gaseous tritium source and a main spectrometer 23 m long and 10 m in diameter. KATRIN published 1.1 eV in 2019, 0.8 eV in 2022, the first direct limit below one electronvolt2, and 0.45 eV in April 2025, based on 259 days of data1. All are upper limits at 90 % confidence; the measured value is consistent with zero.
KATRIN finished its main neutrino-mass campaign after about 1,000 measurement days in 2025 and expects a final sensitivity near 0.3 eV. To go further, Project 8 is developing cyclotron radiation emission spectroscopy, which measures the tiny radio signal an electron emits while circling in a magnetic field. A first tritium measurement in 2023 yielded a limit of about 155 eV5, a proof of principle rather than a competitive value. Experiments with holmium-163 pursue a complementary route using electron capture.
Cosmological bounds
Relic neutrinos from the Big Bang were relativistic when the first structures formed and streamed out of small density lumps. This free streaming slightly suppresses the growth of structure on small scales, by an amount that depends on the sum of neutrino masses Σmν. The Planck satellite, combining the cosmic microwave background with galaxy surveys, found Σmν < 0.12 eV (95 %)6. In 2025 the Dark Energy Spectroscopic Instrument (DESI) combined its baryon acoustic oscillation data with the microwave background and obtained Σmν < 0.064 eV7. These bounds assume the standard cosmological model, ΛCDM; in extended models they become weaker.
Normal or inverted ordering
The sign of Δm²₃₁ is not yet known. In the normal ordering m₁ < m₂ < m₃; in the inverted ordering m₃ < m₁ < m₂. The two cases imply different minimum values of the mass sum: about 0.06 eV for normal and about 0.10 eV for inverted ordering. The DESI limit thus begins to press against the inverted case from the cosmological side. In the laboratory, JUNO in southern China began data taking in August 2025. It studies reactor antineutrinos 52.5 km from two nuclear power plants and aims to determine the ordering from fine wiggles in the energy spectrum over about six years.
Quiz
Test your knowledge
4 questions on this level · Print as worksheet
Expert
Three observables, three different masses
Neutrino mass enters experiments through three distinct combinations of the mass eigenvalues mᵢ and the PMNS matrix elements Uei. Oscillations fix the splittings; NuFIT 6.0 gives Δm²₂₁ = (7.49 ± 0.19) × 10⁻⁵ eV² and Δm²₃ℓ = +2.534 (+0.025, −0.023) × 10⁻³ eV² for normal ordering without Super-Kamiokande atmospheric data8. The absolute scale is probed by β-decay kinematics (mβ), by cosmology (Σmν) and, if neutrinos are Majorana particles, by neutrinoless double beta decay (mββ, treated in a separate deep dive).
With the lightest neutrino mass m₀ → 0 the splittings imply Σmν ≥ 0.059 eV (normal ordering) and Σmν ≥ 0.10 eV (inverted ordering), and mβ ≳ 0.009 eV or 0.048 eV respectively. In the quasi-degenerate regime m₀ ≫ 0.05 eV all three observables approach m₀, 3m₀ and (for mββ) a phase-dependent fraction of m₀.
Tritium β-spectroscopy
Near the endpoint E₀ ≈ 18.57 keV the differential spectrum of molecular tritium behaves as dΓ/dE ∝ (E₀ − E) [(E₀ − E)² − mβ²]^½, summed over final molecular states of the ³HeT⁺ daughter. The fraction of decays within the last eV is only of order 10⁻¹³, so the problem is one of rate, resolution and systematic control. The MAC-E filter solves the resolution problem by adiabatically converting cyclotron motion into longitudinal motion in a falling magnetic field; the integral spectrum is recorded by stepping the retarding potential. Mainz (mβ < 2.3 eV, 95 % CL, 2005)3 and Troitsk (mβ < 2.05 eV, 95 % CL, 2011)4 established the technique.
KATRIN combines a windowless gaseous molecular tritium source of order 10¹¹ Bq with a 10 m diameter main spectrometer and an energy resolution of about 1 eV at 18.6 keV. The analysis fits mβ², E₀, signal normalisation and background. Results: mβ < 1.1 eV (2019), < 0.8 eV (2022)2 and < 0.45 eV at 90 % CL from 259 measurement days in five campaigns (2019–2021), with best fit mβ² = −0.14 (+0.13, −0.15) eV²1. The dominant limitation is statistical; leading systematics include the background from radon decays and from Rydberg atoms sputtered off the spectrometer walls, source plasma potentials and the molecular final-state distribution. A shifted analysing plane reduced the background volume from 2021 onwards. The full data set of about 1,000 days, completed in 2025, targets a sensitivity near 0.3 eV.
Reaching the inverted-ordering floor mβ ≈ 0.05 eV requires a different technology. Molecular tritium imposes an irreducible smearing from rotational-vibrational final states of about 0.4 eV standard deviation, and spectrometer scaling becomes impractical. Project 8 therefore pursues CRES with atomic tritium in a magnetic trap; its first tritium endpoint measurement in a small volume gave mβ < 155 eV (Bayesian)5. Calorimetric electron-capture experiments on holmium-163 (ECHo, HOLMES) offer independent systematics.
Cosmological mass sum
Massive relic neutrinos contribute Ων h² = Σmν / 93.14 eV to the present energy density. After becoming non-relativistic at redshift z ≈ 1900 (m/1 eV) they cluster only above the free-streaming scale, suppressing the matter power spectrum below it by roughly ΔP/P ≈ −8 fν. Constraints come mainly from CMB lensing and the late-time expansion history measured by baryon acoustic oscillations. Planck 2018 with BAO gave Σmν < 0.12 eV (95 %)6. DESI DR2 BAO combined with CMB yields Σmν < 0.0642 eV (95 %, Bayesian, ΛCDM, prior Σmν > 0)7. The posterior peaks at the prior boundary; profile-likelihood analyses extrapolated into an unphysical effective negative-mass region place the minimum at Σmν ≈ −0.04 eV. This is in tension with the oscillation floor at the level of 2–3σ depending on the data combination. Whether this reflects systematics, the preference for dynamical dark energy seen in the same data, or genuine new physics is an open question. In w₀wa cosmologies the bound relaxes to roughly 0.16 eV.
Mass ordering
The mass ordering is accessible through matter effects in long-baseline and atmospheric oscillations and through interference between Δm²₃₁ and Δm²₃₂ in medium-baseline reactor oscillations. In NuFIT 6.0 the inverted ordering is disfavoured by Δχ² = 6.1 when Super-Kamiokande atmospheric data are included, but only Δχ² = 0.6 in the opposite direction without them8. The 2025 joint T2K–NOvA fit showed no strong preference. JUNO, with 20 kt of liquid scintillator at 52.5 km and an energy resolution near 3 % at 1 MeV, aims at about 3σ in roughly six years; its first 59 days already measured sin²θ₁₂ = 0.3092 ± 0.0087 and Δm²₂₁ = (7.50 ± 0.12) × 10⁻⁵ eV². Combined with atmospheric, IceCube–DeepCore, KM3NeT/ORCA and later DUNE data, a determination at 5σ is expected within roughly a decade.
Consistency as the real test
A detection of mβ by KATRIN's final data set would, given Σmν ≈ 3mβ in the degenerate regime, contradict ΛCDM cosmology outright. Conversely, a null result in the laboratory plus a positive cosmological detection would support the standard picture. The three probes are complementary: β-decay is model-independent, cosmology is the most sensitive but model-dependent, and double beta decay additionally tests lepton-number violation.
Quiz
Test your knowledge
4 questions on this level · Print as worksheet
Open questions
- The absolute mass scale: KATRIN's final data set reaches about 0.3 eV; covering the inverted-ordering floor near 0.05 eV needs atomic-tritium CRES.
- The mass ordering, expected from JUNO, atmospheric and long-baseline experiments within about a decade.
- Whether the cosmological preference for very small or effectively negative Σmν points to systematics, dynamical dark energy or new neutrino physics.
Terms to explore
- Neutrino mass
- Absolute neutrino mass scale
- Effective Electron-neutrino mass
- Sum of neutrino masses
- Neutrino mass ordering
- KATRIN
- Tritium beta endpoint method
- Cosmological neutrino mass bounds
- JUNO
- MAC-E filter spectrometer
People
- Enrico Fermi
- Christian Weinheimer
- Vladimir Mikhailovich Lobashev
- Ernst-Wilhelm Otten
- Guido Drexlin
- Joseph A. Formaggio
On the timeline
- 2005Mainz sets a 2.3 eV neutrino-mass limit
- 2011Troitsk publishes a 2.05 eV mass limit
- 2018KATRIN is inaugurated in Karlsruhe
- 2019KATRIN's first limit: below 1.1 eV
- 2022KATRIN breaks the 1 eV barrier
- 2025KATRIN: neutrinos weigh less than 0.45 eV
- 2025KATRIN completes 1,000 days of mass measurements
- 2015Project 8 detects single electrons by their radiation
- 2025JUNO begins data taking
- 2025JUNO's first results set precision records
References
- M. Aker et al. (KATRIN Collaboration), Direct neutrino-mass measurement based on 259 days of KATRIN data, Science 388, 180 (2025), arXiv:2406.13516, doi:10.1126/science.adq9592
- M. Aker et al. (KATRIN Collaboration), Direct neutrino-mass measurement with sub-electronvolt sensitivity, Nature Physics 18, 160 (2022), doi:10.1038/s41567-021-01463-1
- Ch. Kraus et al., Final results from phase II of the Mainz neutrino mass search in tritium β decay, Eur. Phys. J. C 40, 447 (2005), doi:10.1140/epjc/s2005-02139-7
- V. N. Aseev et al., Upper limit on the electron antineutrino mass from the Troitsk experiment, Phys. Rev. D 84, 112003 (2011), doi:10.1103/PhysRevD.84.112003
- A. Ashtari Esfahani et al. (Project 8 Collaboration), Tritium beta spectrum measurement and neutrino mass limit from cyclotron radiation emission spectroscopy, Phys. Rev. Lett. 131, 102502 (2023), doi:10.1103/PhysRevLett.131.102502
- Planck Collaboration, Planck 2018 results. VI. Cosmological parameters, Astron. Astrophys. 641, A6 (2020), doi:10.1051/0004-6361/201833910
- W. Elbers et al. (DESI Collaboration), Constraints on neutrino physics from DESI DR2 BAO and DR1 full shape, arXiv:2503.14744 (2025), arXiv:2503.14744
- I. Esteban, M. C. Gonzalez-Garcia, M. Maltoni, T. Schwetz, A. Zhou, NuFit-6.0: updated global analysis of three-flavor neutrino oscillations, JHEP 12, 216 (2024), doi:10.1007/JHEP12(2024)216
Last updated: · Editorial standard
Cite this page
Text
Neutrino Atlas (2026). “How heavy is a neutrino?”. Retrieved 22 November 2026, from https://neutrino-atlas.com/learn/how-heavy-is-a-neutrino/BibTeX
@misc{neutrinoatlas2026learnhowheavyisaneutrino,
title = {How heavy is a neutrino?},
author = {{Neutrino Atlas}},
year = {2026},
url = {https://neutrino-atlas.com/learn/how-heavy-is-a-neutrino/},
note = {Accessed 2026-11-22},
language = {english}
}