Ask an incandescent lamp when it dies and the answer is simple: the filament breaks, the light stops, the lifetime rating is the point where half of a sample has failed. Ask the same question about an LED and there is no clean death. An LED package loses brightness gradually as its phosphor and semiconductor age, and in a well-designed fixture it usually never stops emitting at all — the electronics around it typically fail first. Lifetime for LEDs is therefore defined not as failure but as decay: the point at which the light output falls to a stated percentage of its initial value.
That definition produces the ratings L70, L80 and L90 that appear across datasheets, and the surrounding figures B10 and B50 that most marketing pages quietly omit. Understanding what each number is — and is not — is one of the highest-leverage skills in fixture procurement, because a 100,000-hour claim and a 35,000-hour claim can describe the same product with equal honesty.
What L70, L80 and L90 actually measure
L70 means the elapsed time at which the LED's light output is projected to reach 70 percent of its initial output — the point conventionally treated as end of useful life for outdoor and industrial applications, where a 30 percent drop is usually noticeable against a maintenance target. L80 marks the 80 percent threshold and is common in indoor commercial and office specifications; L90, the 90 percent threshold, appears in demanding retail and healthcare contexts where light depreciation is visible on merchandise or task surfaces.
Two properties of these ratings matter for interpretation:
- They are projections, not measurements. Quality LED packages degrade so slowly that nobody can wait a decade for the test to finish. Instead, a sample is aged for a fixed period — typically 6,000 to 10,000 hours — and the decay curve is extrapolated by a standardized method with a hard cap on how far the projection may reach.
- They are temperature-dependent. An LED driven at a low junction temperature might take three times longer to reach L70 than the same chip pushed hot. Every lifetime number is meaningless until the temperature condition is attached to it.
This is why two datasheets both reading "L70 50,000 h" can describe very different products: the conditions behind the number are the number.
B10, B50 and the statistics buyers skip
The L-value describes how far the light decays; the B-value describes how much of the population gets there. L70B50 — often abbreviated to just L70 — says that at the stated hour count, 50 percent of units are projected to still deliver at least 70 percent of initial flux. L70B10 is stricter: 90 percent of units still above 70 percent. When a tender specifies L80B10 at 60,000 hours, it is demanding both slow decay and tight uniformity across the population, which is a materially different (and costlier) LED bin and driver quality than L70B50 at the same hours.
For portfolio decisions, the practical reading is: B50 claims are marketing baselines, B10 claims are engineering claims. Distribution programs built on B50 figures will show visible non-uniformity on a lit road or warehouse aisle years before the projected end of life — some fixtures dim, neighbors do not, and the installation looks patchy even though every unit technically still works.
How the numbers are produced: LM-80 and TM-21
The measurement chain has two standardized links. LM-80, maintained by IES, defines how the LED package or module itself is tested: samples run at fixed case temperatures (commonly 55 °C, 85 °C and a third temperature chosen by the manufacturer) for at least 6,000 hours, with photometric and color measurements at intervals. The report belongs to the LED manufacturer — a chip maker such as those in the leading package brands publishes LM-80 files for each product family.
TM-21, also from IES, defines how an LM-80 data set may be legitimately extrapolated: which curve fit to use, and a projection cap of 6 times the test duration (so a 10,000-hour test supports claims up to 60,000 hours, and anything beyond that has no standardized basis). Fixture makers then apply an Arrhenius-type temperature correction to translate the package data to the temperatures their luminaire actually runs at. The full chain — package LM-80, TM-21 projection, fixture thermal design — is what turns "50,000 hours" from a slogan into a defensible statement. Buyers auditing supplier claims can treat the presence of both reports, with matching temperatures, as the dividing line between verifiable and not; our lighting certification guide covers where these reports sit among the other destination-market documents.
Typical rating levels by application are summarized below. Values are conventional industry practice, stated as illustrative.
| Application class | Typical claimed lifetime | Common level | What it implies for the buyer |
|---|---|---|---|
| Outdoor road, industrial, high-burn-hour sites | 50,000-100,000 h | L70, B50 or B10 | 30 percent decay accepted; check B-level for uniformity on multi-fixture sites |
| Office, education, retail general lighting | 50,000-60,000 h | L80 | Decay must stay subtle; color shift over life matters as much as flux |
| High-end retail, gallery, healthcare | 40,000-50,000 h | L90 | Tight binning and premium packages; expect a visible price step |
| Decorative / low-burn-hour residential lines | 25,000-40,000 h | L70B50 | Burn hours are low, so calendar life usually outlasts the claim anyway |
Conventional industry ranges, shown as illustrative orientation — actual claims follow each manufacturer's LM-80 and TM-21 documentation.
Why the fixture matters more than the chip
A common buyer error is treating lifetime as a property of the LED package alone. In reality the package's rated decay is only an input; the fixture's thermal engineering decides how much of it survives. Two fixtures using the identical LED can differ by a factor of two in realized lifetime if one runs the junction 20 °C hotter — through a smaller heat sink, a hotter driver compartment, or sealed construction with no convection path. Driver lifetime interacts the same way: electrolytic capacitor aging is sharply temperature-driven, and on many returned fixtures the driver, not the LED, is what fell short.
When comparing quotations at different price points, the questions that actually separate the tiers are: which LED package and driver brands are on the bill of materials; what case temperature the TM-21 projection assumes; and whether the housing was designed for the claimed hours or for the claimed price. The same discipline applies across categories — it is the reason our outdoor lighting range documents battery and cell grade on solar products, and why fixture-level testing matters as much as package-level ratings on mains products.
Common mistakes when reading lifetime figures
- Treating hours as a warranty. A 100,000-hour projection and a 5-year warranty are different instruments covering different things; one is a statistical decay estimate, the other a contractual remedy. Read them together — see our warranty policy guide for what a 2-5 year term should actually cover.
- Comparing L-values without temperatures. L70 at an unspecified temperature is not a specification. Ask what case or junction temperature the projection assumes.
- Ignoring the B-level. On any installation with more than a handful of fixtures, population uniformity is what the eye sees. B10 versus B50 is the difference between an even field and a patchy one.
- Believing projections beyond the TM-21 cap. A claim of 200,000 hours from a 10,000-hour test has no standardized basis. It is not necessarily false — it is unverifiable.
- Forgetting lumen maintenance in layouts. Lighting calculations should use maintained illuminance — initial lumens multiplied by the maintenance factor at the design hour count — not catalog initial output. Roads designed at initial lumens go dark relative to standard well before L70.
Lifetime ratings reward buyers who read the fine print twice: once for the decay level, once for the conditions behind it. Ask for LM-80 and TM-21 by name, check the temperatures, and the 50,000-hour claim becomes something you can hold a supplier to — via certified partner factories, with reports verified per order.