The first version of the netCDF file format, designed in 1988, is still supported by current software. Other netCDF format variants have been created as needed: 64-bit offset, netCDF-4, netCDF-4 classic model. CDL and NcML are text representations.

NetCDF representations:

• classic — the only format for netCDF data created between 1989 and 2004 by various versions of the reference software from Unidata. Supports the classic data model. Recently added: support for Unicode names.
• 64-bit offset — introduced in late 2004 with version 3.6 for creation of and access to much larger files. Supports only the classic data model. Works fine on 32-bit platforms. Very similar to the classic format except for use of 64-bit file offsets for start of variables. Still has size restrictions, for example 4 GiB per record for variables.
• netCDF-4 — implementation of enhanced netCDF API using enhanced HDF5 as a storage layer. Requires HDF5 1.8 or later (April 2008). Supports both classic and enhanced data model. Adds performance features as well as support for enhanced netCDF model.
• netCDF-4 classic model — variant of HDF5-based netCDF-4 format that supports only classic data model. Supports performance benefits while preserving API compatibility with netCDF-3 software.
• CDL — "Common Data Language", since 1989, the original ASCII representation for netCDF data, output from ncdump, input for ncgen. Recently updated to permit escaped special characters in names and Unicode names for both netCDF-3 and netCDF-4. Updated syntax fully supports enhanced data model.
• NcML — XML dialect created in 2002 for representing netCDF data, as well as virtual netCDF datasets with additional metadata or aggregations of other datasets. The "-x" option of ncdump specifies NcML output. Java netCDF can read NcML virtual datasets. Syntax for representing enhanced data model still evolving.

2008 Unidata NetCDF Workshop for Developers and Data Providers > Formats and Performance