Blender features an internal file system that allows one to pack multiple scenes into a single file (called a ".blend" file).

The actual ".blend" file is similar to the EA Interchange File Format, starting with its own header (for example BLENDER_v248) that specifies the version, endianness and pointer size, followed by a collection of binary chunks storing the data blocks, and all the type and struct definitions also known as DNA. Although it is hard to read and convert a ".blend" file to another format using external tools, the readblend utility can do this. Dozens of import/export scripts that run inside Blender itself, accessing the object data via API, make it possible to inter-operate with other 3D tools.

Jeroen Bakker documented the Blender file format to allow inter-operation with other tooling. The document can be found at the The mystery of the blend website.^[14] A DNA structure browser^[15] is also available on this site.

Blender organizes data as various kinds of "data blocks", such as Objects, Meshes, Lamps, Scenes, Materials, Images and so on. An object in Blender consists of multiple data blocks - for example, a polygon mesh has at least an Object and Mesh data block, and usually also a Material. This allows various data blocks to refer to each other; there may be, for example, multiple Objects that refer to the same Mesh, allowing the mesh to be duplicated while only keeping one copy of the mesh data in memory, and allowing subsequent editing of all duplicated meshes at the same time. Data block relationships can also be changed manually. Data blocks can also be referred to in other .blend files, allowing the use of .blend files as reusable object libraries.