CD-ROM/XA extends CD-ROM by adding some of the CD-i disc features, such as using mode 2 tracks with interleaved compressed-audio and other data. Thus, CD-ROM/XA (eXtended Architecture) is often called the bridge format between CD-ROM and CD-i, though the relationship is mostly the sector types. CD-ROM/XA applications still require specific code for each target platform.

CD-ROM/XA specifies a bridge format so that a CD-ROM/XA disc can be read on a CD-i player as well as on a computer under MSCDEX, provided that you use a conforming (.sys) driver. On a normal yellow-book CD-ROM, a data track contains only mode 1 data sectors. On a CD-ROM/XA, a data track contains only mode 2 data sectors. A mode 2 sector can be of two formats, form 1 and form 2. A form 1 sector contains 2,048 bytes of user data, together with Error Detection (EDC) and Error Correction (ECC) bytes. A form 2 sector contains 2,324 bytes of raw data, for example, ADPCM audio or video data. If the device driver delivers 2,048 bytes of data when MSCDEX asks for it, regardless of whether the sector is mode 1 or mode 2/form 1, the computer is CD-ROM/XA compatible.

CD-ROM/XA discs contain mode 2 sectors and were designed to allow audio and other data to be interleaved and read simultaneously. This avoids the need to load images first, then play CD audio tracks.

The CD-ROM/XA specification also defines certain image and audio formats to use:

• The graphics formats include 256-color modes that are compatible with computer formats and CD-i.
• The audio used is Adaptive Delta Pulse Code Modulation (ADPCM), which is also defined for CD-i.

This CD-ROM/XA format has not been successful in itself, but there are three important formats based on it: Photo CD, Video CD, and CD EXTRA. CD-i also uses mode 2 sectors and shares some commonality with CD-ROM/XA, for example, both use ADPCM audio.

Mode 2 sectors are used for those formats based on CD-ROM/XA and can be either form 1 or form 2. Form 1 sectors contain 2,048 bytes with both ECC and EDC, while form 2 sectors contain 2,324 bytes of user data per sector and EDC. Form 2 sectors, with no error correction, are suitable only for data where errors can be concealed, for example, audio or video data.