How Long Do Burned CDs and DVDs Last?

Two discs that look identical

Put a film you bought on DVD next to a disc you burned yourself. They are the same size, the same weight, and they play in the same drive. They are not the same object, and they do not age the same way.

The pressed disc — the one that arrived in a case with printing on it — was made in a molding machine. In the words of the standard reference on the subject, Care and Handling of CDs and DVDs, published by the Council on Library and Information Resources with NIST in 2003, a stamper “impress[es] the pits (depressions) and lands (surface), which form the data, into the polycarbonate substrate surface.” The data is a physical shape molded into plastic.

The disc you burned has no pits. It has a layer of organic dye, and your drive’s laser wrote to it by causing “a chemical change” in that dye. The same document is direct about the consequence: the dye “degrades naturally but slowly over time,” and “high temperatures and humidity will accelerate the process.”

The numbers, and who actually said them

Search for how long a CD-R lasts and you will be told 100 years, often attributed to NIST. The attribution is the problem.

Here is what the 2003 NIST and CLIR guide actually says. Among manufacturers that have done testing, “there is consensus that, under recommended storage conditions, CD-R, DVD-R, and DVD+R discs should have a life expectancy of 100 to 200 years or more,” while rewritable discs — CD-RW, DVD-RW, DVD+RW, DVD-RAM — “should have a life expectancy of 25 years or more.” Pressed discs are put at 20 to 100 years, with what the authors call “an increased level of uncertainty.”

Read that first clause again. It is a consensus among manufacturers, reported by NIST and CLIR. It is not a measurement by NIST or CLIR. And the very next sentence in the document is the one nobody quotes:

The guide then reports the one independent number it has. “An accelerated aging study at NIST estimated the life expectancy of one type of DVD-R for authoring disc to be 30 years if stored at 25 °C (77 °F) and 50% relative humidity.” The authors immediately add that “this testing for R discs is in the preliminary stages.”

One type of disc, one preliminary study, twenty-odd years ago. That is the entire independent evidence base, and it is a long way from a hundred years. It also happens to mean that a disc burned in 1998 is now past the only figure anyone actually measured.

There is a second gap worth knowing about. Manufacturers claim blank discs have a shelf life of five to ten years before you record on them, but the same guide notes that no expiration dates are printed on them, “nor are there published reports of tests to verify these claims.” If you burned a disc in 2006 from a spindle bought in 2001, nobody can tell you what you started with.

What failure looks like

The pressed-disc failure mode is the one people can see. Longevity there “is determined by the extent to which its aluminum layer is exposed to oxygen,” and once moisture and oxygen reach it, the aluminum “loses its reflectivity” and becomes, in the guide’s memorable phrase, “oxide-dull and much less reflective, like the color of a typical aluminum ladder.” Oxygen gets in most easily through “scratches, cracks, or delaminated areas in the label.” Hold the disc up to a light and you may see pinholes or a bronzing at the edge.

Recordable discs use gold, silver or a silver alloy rather than aluminum. Gold does not corrode, which is why gold-layer archival discs were sold at a premium. Silver does: the guide notes it is “susceptible to corrosion if exposed to sulfur dioxide” — which is to say, to polluted air.

But the failure that catches people out on their own burned discs has no visual signature at all. The dye simply drifts, error correction absorbs the damage silently for years, and then one day it cannot. A disc that reads perfectly today and fails next year will look identical on both days. This is the practical reason not to treat a shelf of discs as a backup you have: until you have put one in a drive and read it end to end, you have a shelf of discs you hope you have.

You will see this described online as “disc rot.” The mechanism is real and documented. The phrase is a consumer coinage — it appears in no standards document we could find, so treat it as shorthand rather than a technical term.

The habits that shortened their lives

Two things nearly everybody did in 2004 are specifically advised against by the people who preserve media for a living.

• Adhesive labels. The guide is unambiguous: adhesive labels “should not be applied to optical discs destined for long-term storage (more than five years),” because “the label could delaminate over time.” Recall that delaminated label areas are one of the routes oxygen takes to the reflective layer. The sticker meant to help you identify the disc is a documented failure path.
• Storing them flat. “Store discs upright (book style),” the guide says, and do not store them horizontally for years, because “long-term horizontal storage, particularly in a heated environment, can cause the disc to become permanently bowed.” A stack of discs in a loft is both of those conditions at once.

On marking discs, the recommendation is a non-solvent-based felt-tip permanent marker, and never a fine-point or rolling-ball pen, which “may scratch or depress the surface.” It is worth noting that the guide is honest about the limits of its own advice here: on the question of what marker solvents actually do, it states there are “no explicit lab test results.”

The published storage conditions are more achievable than they sound. The guide recommends below 20 °C but above 4 °C, at 20% to 50% relative humidity, and suggests 18 °C and 40% RH as suitable for long-term storage. It also reproduces figures from other bodies: the US National Archives and Records Administration (2001) at 20 °C and 40% RH, ISO 18925 (2002) at −10 to 23 °C and 20–50% RH, and the National Library of Canada (1996) at 15–20 °C and 25–45% RH. They broadly agree, and what they describe is an ordinary heated room — not a loft, not a garage, and not a conservatory.

What to move them onto

The reason to read a shelf of discs now rather than next year is simply that the drives are disappearing along with the discs. Few PCs have shipped with an optical drive for years, and an external USB one is the kind of thing that is easy to buy today and awkward to find in a decade.

Where the files should land afterwards deserves the same skepticism this article has applied to discs. Here is what the specifications actually say, and where they go quiet:

That last row is the most useful one in the table, and it is why the sensible conclusion is not “copy the discs onto one external drive and put it back in the loft.” That is the same mistake with newer plastic. What replaces a disc is not another single object; it is a working copy you actually use plus copies elsewhere, on a schedule — the reasoning behind 3-2-1, and the reason a shelf that has never been read does not count as one of the three.

Four things you will read that are not true

Each of these circulates widely, and each is a misreading of a real document.

1. “NIST says CD-Rs last 100 to 200 years.” NIST and CLIR reported that figure as a consensus among manufacturers who had done their own testing, and in the same section said that few, if any, independent reports exist. NIST’s own preliminary test on a DVD-R produced 30 years.
2. “The US military proved M-DISC lasts 1,000 years.” The 2009 US Navy report from China Lake is real, and its finding is genuinely striking: “None of the Millenniata media suffered any data degradation at all. Every other brand tested showed large increases in data errors after the stress period.” But that report contains no lifespan figure whatsoever. The 1,000-year number comes from the manufacturer’s own 2013 press release about testing it commissioned. The Navy report also discloses that the test protocol was “created by Millenniata” and that “the irradiance levels during light exposure could have been controlled better.” A strong result under a vendor-authored protocol is worth knowing; it is not a millennium.
3. “An unplugged SSD loses your data in a year.” This comes from JEDEC’s JESD218, which requires a consumer-class drive to retain data for one year powered off at 30 °C. What gets dropped in the retelling is that this is a floor the drive must still meet after being written to 100% of its rated endurance — the end of its working life, not the state of a lightly used drive in a drawer. The enterprise figure of three months is not worse engineering either: enterprise drives are qualified at a hotter 40 °C power-off temperature.
4. “Hard drives last three to five years in storage.” We could not reach a primary source for this at all. It may well be reasonable, but nobody authoritative appears to have published it, so we are not going to repeat it as though someone had.

The part specific to photographs

There is a failure specific to photo archives that none of the preservation literature covers, because it is not about the medium at all.

Get the files off the discs and you will very often find them arriving in a single undifferentiated heap, ordered by the day the disc was burned or the day you copied them — not by when the photographs were taken. A disc burned one evening in 2006 might hold three years of family life, and after the rescue it all shares one date. The pixels came through perfectly. The chronology did not.

Sometimes the real capture date is still there, sitting in the file’s metadata, and only the sorting is wrong. Sometimes it is genuinely gone — scans of prints never had one to begin with. The two cases need different answers, and the difference matters more than any single tool: evidence that exists can be recovered, and evidence that does not exist should be admitted rather than invented. That distinction is what our walkthrough on fixing wrong photo dates is built around.