Digital obsolescence is the risk of data loss because of inabilities to access digital assets, due to the hardware or software required for information retrieval being repeatedly replaced by newer devices and systems, resulting in increasingly incompatible formats. While the threat of an eventual "digital dark age" (where large swaths of important cultural and intellectual information stored on archaic formats becomes irretrievably lost) was initially met with little concern until the 1990s, modern digital preservation efforts in the information and archival fields have implemented protocols and strategies such as data migration and technical audits, while the salvage and emulation of antiquated hardware and software address digital obsolescence to limit the potential damage to long-term information access.

A false sense of security persists regarding digital documents: because an infinite number of identical copies can be created from original files, many users assume that their documents have a virtually indefinite shelf life. In reality, the mediums utilized for digital information storage and access present unique preservation challenges compared to many of the physical formats traditionally handled by archives and libraries. Paper materials and printed media migrated to film-based microform, for example, can be accessible for centuries if created and maintained under ideal conditions, compared to mere decades of physical stability offered by magnetic tape and disk or optical formats. Therefore, digital media have more urgent preservation concerns than the gradual change in written or spoken language experienced with the printed word.

Little professional thought in the fields of library and archival science was directed toward the topic of digital obsolescence as the use of computerized systems grew more widespread and commonplace, but much discussion began to emerge in the 1990s. Despite this, few options were proposed as genuine alternatives to the standard method of continuously migrating data to increasingly newer storage media, employed since magnetic tape began succeeding paper punch cards as practical data storage in the 1960s and 1970s. These basic migration practices persist into the modern era of hard disk and solid-state drives as research has shown many digital storage mediums frequently last considerably shorter in the field compared to manufacturer claims or laboratory testing, leading to the facetious observation that "digital documents last forever—or five years, whichever comes first."

The causes for digital obsolescence are not always purely technical. Capitalistic accumulation and consumerism have been labeled key motivators toward digital obsolescence in society, with newly introduced products frequently assigned greater value than older products. Digital preservation relies on the continuous maintenance and usage of hardware and software formats, which the threat of obsolescence can interfere with. Four types of digital obsolescence exist in the realm of hardware and software access:.

Because the majority of digital information relies on two factors for curation and retrieval, it is important to separately classify how digital obsolescence impacts digital preservation through both hardware and software mediums.

Hardware concerns are two-fold in archival and library fields: in addition to the physical storage medium of magnetic tape, optical disc, or solid-state computer memory, a separate electronic device is often required for information access. And while proper storage can help mitigate some environmental vulnerabilities to storage formats (including dust, humidity, radiation, and temperature) and extend preservation for decades, there are other inevitable endangering factors. Magnetic tape and floppy disks are vulnerable to both the deterioration of adhesive holding the magnetic data layer to its backing or the demagnetization of the data layer, commonly called "bit rot"; optical discs are specifically susceptible to physical damage to their readable surface, and to oxidation occurring between improperly sealed outer layers; a process referred to as "disc rot" or, inaccurately, "laser rot" (particularly in reference to LaserDiscs). Older forms of floating-gate MOSFET based read-only-memory storage such as (some) cartridges and (most) memory cards encounter their own form of bit rot when the charges representing individual bits of binary information dissipate beyond a certain level (called "flipping") and the data is rendered unreadable.

The operability of a format’s appropriate playback or recording device possess their own vulnerabilities. Cassette decks and disk drives rely on the functionality of precision-manufactured moving parts that are susceptible to damages caused by repetitive physical stress and foreign materials like dust and grime. Routine maintenance, calibrations, and cleaning operations can help extend the lifetime of many devices, but broken or failing parts will need repair or replacement: sourcing parts becomes more difficult and expensive as the supply stock for older machines reaches scarcity, and user technical skills grow challenged as newer machines and storage formats use less electromechanical parts and more integrated circuits and other complex components.

Only a decade after the 1970s Viking program, NASA personnel discovered that much of the mission data stored on magnetic tapes, including over 3000 unprocessed images of the Martian surface transmitted by the two Viking probes, was inaccessible due to a multitude of factors. While in possession of indecipherable notes written by long-departed or deceased programmers, the computer hardware and source code needed to correctly run the decoding software had been replaced and disposed of by the agency. Information was eventually recovered after more than a year of reverse engineering how the raw data was encoded onto the tapes, which included consulting with the original engineers of the Viking landers’ cameras and imaging hardware. NASA experienced similar issues when attempting to recover and process images from 1960s lunar orbiter missions. Engineers at the Jet Propulsion Laboratory acknowledged in 1990, following a one-year search that located a compatible data tape reader at a United States Air Force base, that a missing part might need to be rebuilt in-house if a replacement could not be sourced from computer salvage yards.