Aside from low cost per gigabyte of storage (at which CDR has been eclipsed by at least two other technologies), the main selling point of CDR from the beginning has been its potential for long term archival storage. Given the fact that the dyes used in CDR were chosen for their long term stability and the layers were carefully chosen to avoid interaction and were properly sealed (a lesson learned for the "rot" of pressed music CDs in the UK), the claims by CDR makers of 70, 100 and even 200 years seemed quite reasonable.

Indeed, formulas designed to determine the lifetime of the media suggested they would last hundreds, even thousands of years, but they partially relied on factors not fully known. Interesting enough, one of the factors was the speed at which it was written, more on that later.

Despite the general sense of optimism some questioned the archival stability of CDR as early as 1999. Some computer and consumer audio publications had even done tests. These test were not based on actual aging, of course, since CDRs hadn't been around long enough for that. Instead they involved repeated heating and cooling, high humidity and drying to do "accelerated aging" tests. While such tests are considered valid for cars, showing years of opening and closing doors, for example, for CDRs all they really show is how well they hold up to concentrated abuse.

Since some CDRs failed early and others held up for the duration of the tests, this may show the relative merits of the tested CDRs, but it says very little if anything about how well they will hold up under true archival conditions.

Of course, the relevance of tests done on CDRs in the late 90s to the same brands of DVDRs in 2007 is questionable at best, especially since the manufacturing plants used then aren't necessarily used today. Other studies have indicated an actual archival life of 15 to 20 years, certainly long enough to migrate important data to a newer storage medium. More pessimistic estimates of 7 years or less have already been proven false by actual experience.

Another question would be if a CDR that begins failing after 15 years would suddenly become unreadable or if it would just show a few errors in some of the files, much like magnetic storage media such as floppies and data tapes. If it is the latter, having multiple archival copies of important data would be a good defense.

The former scenario brings up an issue that is much more troubling to some users: Anecdotal accounts of complete failure in a much shorter time.

One user, for example, wrote about "a dozen CDRs kept for 3 years under archival conditions" which were found to be "stone cold dead". Such an account is anecdotal and obviously not repeatable as an experiment. The fact that I have many CDRs like the one shown on this page which are readable without errors after 9+ years and have never lost data on a CD for any reason other than actual physical destruction (accidental or otherwise!), causes one to wonder what I am doing right and what he did wrong.

Could it be the writing speed? Since he wrote them in 2003 they may have been written at 32X+, and the formula does show an archival life inversely proportional to the writing speed. [assuming a writing speed of at least 2X since 1X tends to "overburn", kinda like putting too much ink on a stamp pad]. It seems unlikely, however, that even a 48X writing speed would cause failure of actually written data in such a short period of time!

What type of CDR was it? Though some users have reported problems only with one particular type or brand of CDR, I have used a variety of CDRs and though some showed more errors in burning, none showed problems after a verified write.

What labelling method (if any) was used on the disk? This is important because many users have reported that the only disks they had problems with were ones they attached a stick on label to 2+ years ago. The same users report no problem with disks labelled with a black marker, suggesting that sometimes the crudest methods are also the best! I haven't had enough experience with adhesive labels to comment one way or another on this and most of the ones I did use were on disks given out to other people years ago.

continued