How Long Do Stars Live? The Mass-Lifetime Relation

Stars live anywhere from about two million years to spans so long that no star at the small end has ever died of old age. One property sets the whole range: mass. The Sun, a middleweight, gets roughly 10.4 billion years of stable hydrogen burning, and at a measured age of about 4.6 billion years it is a little under halfway through. A thirty-solar-mass star gets about two million years. A half-solar-mass red dwarf gets more than fifty billion — several times the current age of the universe.

The pattern runs opposite to fuel intuition: the more massive the star, the shorter its life. Mass buys fuel at a linear rate, but luminosity — the rate the fuel is spent — climbs far faster than mass, so heavier stars race through bigger tanks in less time. That inversion is the entire answer to how long stars live, and the mass-lifetime relation below makes it quantitative. Every figure in this post can be reproduced, parameter by parameter, with the star lifetime calculator.

The stellar lifetime formula

A star's main-sequence lifetime — the hydrogen-fusing phase that accounts for roughly 90% of its luminous life — is estimated with the nuclear timescale: energy available, divided by the rate it is radiated away.

t = ε f M c² ÷ L — and because luminosity follows the mass-luminosity relation L ∝ M^α, lifetime scales as t ∝ M^(1−α)

Three parameters do all the work. The efficiency ε comes from nuclear physics: fusing four hydrogen atoms into one helium-4 atom converts about 0.7% of their rest mass to energy — AME2020 atomic masses give 0.0071185, roughly 2% of which escapes as neutrinos, and the conventional ε = 0.007 sits between the raw and corrected values. The core fuel fraction f is the share of the star's mass that ever gets hot enough to fuse; the textbook figure is 0.1, because only the core burns. And α is the exponent of the mass-luminosity relation, conventionally 3.5 for mid-range stars. With α = 3.5, lifetime falls as M^−2.5: doubling the mass merely doubles the fuel while raising the burn rate about elevenfold (2^3.5 = 11.31).

Working it for the Sun

Nothing here needs a computer. The Sun's nuclear reservoir is

0.007 × 0.1 × 1.98841 × 10³⁰ kg × 8.98755 × 10¹⁶ m²s⁻² = 1.25097 × 10⁴⁴ joules,

where the last factor is c². Divide by the solar luminosity of 3.828 × 10²⁶ watts:

1.25097 × 10⁴⁴ ÷ 3.828 × 10²⁶ = 3.26794 × 10¹⁷ seconds.

A Julian year is 31,557,600 seconds, so 3.26794 × 10¹⁷ ÷ 3.15576 × 10⁷ = 1.03555 × 10¹⁰ years — 10.355 billion, the origin of every "the Sun will live about ten billion years" you have ever read. Two notes attach. The answer moves in exact proportion to f and ε: swap in the raw AME2020 mass defect of 0.0071185 for 0.007 and the Sun gains 1.7%, to 10.531 billion years. And this is main-sequence time only — the giant phases that follow are shorter and are not counted here.

Ten solar masses, three hundred times less time

Now a ten-solar-mass star, a hot B-type star of the kind that ends as a supernova. Under α = 3.5 its luminosity is 10^3.5 = 1000 × √10 = 3,162.28 times the Sun's. It carries ten times the fuel and spends it 3,162 times faster, so it gets 10 ÷ 3,162.28 = 1/316.23 of the Sun's time: 10.355 billion ÷ 316.23 = 32.75 million years.

Push to thirty solar masses: 30^2.5 = 900 × 5.4772 = 4,929.5, so the lifetime is 10.355 billion ÷ 4,929.5 ≈ 2.1 million years. This is why the most massive stars in the sky are always young — two million years is nothing on a galactic timescale, so any such star still shining formed essentially yesterday.

The lifetime ladder

The top row deserves a pause. 0.5^−2.5 = 2^2.5 = √32 = 5.657, so a half-solar-mass star lives 10.355 × 5.657 = 58.6 billion years — and the universe is 13.787 billion years old. Every red dwarf that has ever formed is still quietly on the main sequence; not one has yet died of old age. The 2 M☉ row comes the same way (10.355 ÷ 5.657 = 1.83), and every row assumes f = 0.1, ε = 0.007 and a single exponent of 3.5 — which is where the honest trouble starts.

The exponent is a fit, not a law

L ∝ M^3.5 is an empirical compromise, not handed-down physics. The best modern calibration — Eker et al. (2018), fitted to 509 stars in detached eclipsing binaries, the only systems whose masses are measured directly — finds the exponent changes with mass: 2.028 from 0.179 to 0.45 M☉, 4.572 up to 0.72, a steep 5.743 up to 1.05, then 4.329 up to 2.40, 3.967 up to 7, and 2.865 from 7 to 31 M☉. The single 3.5 is usually quoted for roughly 2–20 M☉ and is stretched everywhere else.

That choice moves answers a lot. Hipparcos astrometry dates the Hyades cluster at 625 ± 50 million years (Perryman et al. 1998). Ask which mass lives exactly 0.625 billion years and the inverse formula returns 3.074 M☉ with the textbook α = 3.5, 2.549 M☉ with α = 4.0, and 2.324 M☉ with Eker's measured 4.329 for that mass range — a 24% change in the exponent moved the answer by 32%. The red-dwarf row above roughly doubles under Eker's 4.572 for its bin: 0.5^(1−4.572) = 2^3.572 ≈ 11.9, giving 10.355 × 11.9 ≈ 123 billion years instead of 58.6. The conclusion that survives every exponent is the qualitative one — red dwarfs outlive the present universe several times over.

So treat every absolute lifetime here as good to about one significant figure. The most robust output is the ratio between two lifetimes, (M/M☉)^(1−α), because ε and f cancel out of it completely.

What the formula quietly assumes

• Constant luminosity. Real stars brighten as they burn — the Sun is already about 30% brighter than when it reached the main sequence — so the clock runs faster late in life than this model admits.
• A main sequence exists. Below about 0.08 M☉ hydrogen never ignites; the object is a brown dwarf and has no main-sequence lifetime at all.
• A calibrated mass. Measured mass-luminosity data covers 0.179–31 M☉. Outside that band, every exponent on offer is an extrapolation past its data.
• Populations, not individuals. The Hyades spread above is the demonstration: one power law cannot date a single star. Real cluster ages come from fitting full stellar-evolution models.

Weighing and timing a real star

The formula consumes a mass and a luminosity, and a telescope hands you neither — a star is a point of light at an unknown distance. Luminosity requires distance, and the only fully geometric distance measurement in astronomy is trigonometric parallax: d = 1/p, distance in parsecs from parallax in arcseconds. Proxima Centauri, the nearest star of all, has the largest parallax ever measured, 768.0665 milliarcseconds (Gaia EDR3, via SIMBAD): d = 1000 ÷ 768.0665 = 1.30197 parsecs, which at 3.26156 light-years per parsec is 1.30197 × 3.26156 = 4.2465 light-years. The parallax distance calculator performs that inversion with the catalogue error bar attached, and warns once the parallax uncertainty passes the roughly-20% threshold (Bailer-Jones 2015) beyond which naive inversion stops being trustworthy. Masses come the harder way, from the orbits of binary stars — which is exactly why Eker's calibration sample is eclipsing binaries and nothing else.

That is the full chain behind a sentence like "the Sun is a little under halfway through its life": parallax gives distance, distance turns brightness into luminosity, binary orbits give mass, and mass gives time. Every link in it is arithmetic you can redo by hand, and the astronomy tools on Quanta are deliberately built to show that arithmetic — assumptions, exponents and all — rather than round it away. And if a figure in this post will not reproduce in the calculator with the stated inputs, write to us with the numbers you tried: either the post or the tool is wrong, and at one honest significant figure there is nowhere for the discrepancy to hide.

Sources

• Eker et al. (2018) — main-sequence mass–luminosity relations, MNRAS
• Prša et al. (2016) — IAU 2015 Resolution B3, nominal solar values
• Particle Data Group — astrophysical constants and parameters
• Perryman et al. (1998) — the Hyades: distance, structure, dynamics and age