Table of Contents

Solar panels do not have a single lifetime printed on the glass. They have a warranty, which is a contract, and they have a degradation rate, which is a measurement of how fast power falls in the field. Jordan and Kurtz, in an NREL review, assembled nearly 2,000 published degradation rates from about 40 years of flat-plate module and system tests. The median was about 0.5 percent per year. The mean was about 0.8 percent per year, because the distribution has a long tail of worse modules. About 78 percent of the reported rates were under 1 percent per year. That is a survey of literature, not a random sample of every panel ever sold, and the authors say so. High-degradation modules are less likely to stay in the field and keep being reported.

What “how long solar panels last” is actually asking

If the question is “when do they go to zero,” the literature does not answer with a year. A median loss near half a percent per year is a slow decline, not a cliff. If the question is “when might power fall below a warranty floor,” that depends on the contract and on the module you received. A buyer’s specification such as Xcel Energy’s 2023 rules asks for a power warranty of at least 25 years with no more than 0.5 percent loss per year after year one. That cap happens to sit near the median of the older field survey. It is still a promise of a remedy, not a measurement that your serial number will follow the median. How those contracts are built is in what a solar panel warranty covers.

A 2016 compendium by the same lead author, with a much larger set of rates, put crystalline-silicon medians around 0.5 to 0.6 percent per year and means around 0.8 to 0.9 percent. It also warned that rates measured against the nameplate, rather than against a beginning-of-life flash test, pick up whatever gap existed on day one. Outdoor soiling and measurement noise widen that gap. Hotter climates and mounts that stay hot may raise degradation for some products and not others. None of those sentences is “your roof will be at 90 percent in year twenty.” They are reasons to ask for the flash file and to keep the glass clean before you call a slow decline a failure.

Short-term light-induced degradation is a different clock. The 2012 review tried to keep it out of the long-term histogram. PERC cells have a known light-induced issue, with stabilization methods in the literature. A drop in the first weeks is not the 0.5 percent per year line. The cell background is in what a PERC solar cell is.

What ends a solar panel before degradation does

Cracked cells from someone walking on the glass, a failed diode, water in a junction box, a connector left half-mated, and hail that the certificate did not cover are not the median rate. They are damage. Snow cover is not degradation either. It is weather, and it leaves. Soiling is reversible. A performance warranty measured through dirt measures the dirt. PNNL’s field note that some panel warranties run 25 years and many run 10 is about the paper. Inverters, in that same note, are often warranted closer to 5 years, with extensions to 10 or 15. The array can outlast the first inverter on paper and in practice. Budget the inverter as the part that may be replaced. That is planning, not a prediction of year five.

The module is still live in daylight on the day you decide it is “old.” Age does not make the DC side safe. Shutdown remains the AC disconnect, then the DC disconnect, by a trained person. Recycling at the end is a chemistry-specific path, not a single bin for every wafer and every thin film. The cadmium telluride route is its own article. Do not generalize it.

How to use a degradation rate without lying

Number people quote What it is
About 0.5 percent per year Median of the Jordan and Kurtz literature survey. Not your serial number.
About 0.8 percent per year Mean of that same survey. The tail of poor modules pulls it above the median.
0.5 percent per year after year one A cap in one utility’s buying rules. A contract, not a field study.
25 years A common performance-warranty term, including a US work-specification floor of 80 percent at 25 years for work under that specification. Not a measured life.

If you are estimating payback, name which row you used. A blank “panels last 25 years at full power” uses none of them. The arithmetic is in how a solar payback period is calculated.

Frequently asked questions

Do solar panels last 25 years?

Many performance warranties run 25 years, and a US standard-work specification for covered projects asks for at least 80 percent of nameplate over that period. Field studies show a slow median decline, not a switch-off at year 25. Modules can outlast the paper or fail inside it from damage that is not degradation.

Will my panels lose half a percent a year?

That is the median of a large published set, not a promise. The mean was higher. Hot mounts and some thin-film products have looked worse in later reviews. Your evidence is the flash file now and a proper measurement later, not the median alone.

Does a dirty panel count as wearing out?

No. Soil washes off. Degradation is the change that remains after the module is clean and the measurement is translated to standard test conditions. A hot afternoon reading is not that test.

Sources and references

Median about 0.5 percent per year, mean about 0.8 percent, nearly 2,000 published rates: Jordan and Kurtz, NREL preprint, 2012. Later crystalline-silicon medians around 0.5 to 0.6 percent and the nameplate-reference warning: Jordan and colleagues, 2016 compendium. Warranty cap of 0.5 percent per year after year one in one buyer’s rules: Xcel Energy, 2023. Program floor of 80 percent for 25 years: NLR Standard Work Specifications. Typical contract lengths in the field: Pacific Northwest National Laboratory.