Jef Poskanzer (jef@acme.com) invented the PBM format in the 1980s to provide the computer graphics world with a common, trouble-free format, a lingua-franca, for bitmaps. The format was designed to be simple enough that it could transmitted within an email message without any special encapsulating and survive any translations and recoding that an email message might go through and be easily extractable on the other end.

In 1988, Jef distributed the forerunner of Netpbm, Pbmplus, which gathered together the various tools he had developed to work with PBM files. These were mainly tools to convert between PBM and other existing graphics formats, making it possible to deal with the Tower Of Babel situation that had arisen with the proliferation of graphics formats.

By the end of 1988, Jef had added the PGM and PPM formats and lots more tools to Pbmplus.

In 1991, Jef added a variety of programs and code contributed by the user community, and then stopped maintaining Pbmplus. Jef never formally renounced support for it, but simply didn't get around to distributing any updates for over a year following his final December 10, 1991 release.

In 1993, Netpbm was developed to replace Pbmplus. Netpbm was nothing more than a new release of Pbmplus, and named for the fact that people all over the world would maintain the package by submitting fixes and enhancements over the Net. This was a time when such worldwide collaboration was still novel.

Then Netpbm apparently fell into neglect with the last release by its regular maintainer in March 1994. In September of 1995 Anthony Thyssen released an unofficial update of it called Netpbm-1mar1994.p1, while disclaiming any responsibity for ongoing upkeep.

The documentation in that 1995 release mentioned two different official mailing lists concerning the package, but by April 1999, neither existed anymore.

In September 1999, Bryan Henderson needed a minor bug in Xbmtopbm fixed, and wanted to clean up the mess of unsupported versions and outdated documentation of these tools. After checking around to see if anyone had any claims of ownership of the package, and finding none, Bryan assumed control and responsibility for Netpbm.

In November 1999, Thorbjoern Ravn Anderson did the same thing, for the same reason (apparently unaware of Bryan's work) and made available an updated version (based on the 1994 release) as described by http://sunsite.auc.dk/netpbm. Shortly thereafter, he discontinued his effort in deference to the new Pbmplus effort mentioned below. Bryan did not know about Anderson's work until March 2000.

In August 2000, the PAM format appeared. But it took a long time after that for there to be enough programs and library routines to make it useful. It still is barely known and used; PAM is more of a direction or philosophy than a practical tool. Along with PAM came a whole new suite of library routines for processing both classic PNM and PAM format images. The new "pam" routines are easier to use.

In parallel, and unknown to Bryan, Jef was working on a new release under the original name Pbmplus and claiming again (or maybe still) to be maintaining Pbmplus. He did a limited release of a beta version of it in November 1999. A mailing list pbmplus@acme.com was active. Bryan learned of this effort's existence in March 2000. No widespread distribution ever resulted from this.

Pbmplus and Netpbm were once part of the body of X/Windows contributed software, distributed on ftp.x.org (directory contrib/utilities). Bryan didn't receive a response to an inquiry to the ftp.x.org owner as to the prospect of cleaning up the Netpbm related files there. So Bryan instead made Metalab the new home of the source code. But Metalab had pretty low quality too, and in April 2000 Bryan moved the package's home to Sourceforge.

In June 2002, Bryan reorganized the package. Before, it was organized to reflect the evolution of the package: first PBM only, then PGM was added, then PPM, and then the multi-format PNM programs. There were four separate subroutine libraries and each of the PBM, PGM, PPM, and PNM components was independent of the components invented later. Documentation was in the form of traditional man pages. In Release 10, the package was integrated into a single monolithic package. There was one subroutine library and the source tree was divided according to program function rather than what generation of Netpbm it related to. All documentation was converted to HTML.

In January 2004, Netpbm started to get the linear/gamma-adjusted sample value thing right. Many programs had always used the gamma adjusted values from Netpbm images as if they were linear, which produced incorrect and rather terrible results. Netpbm 10.20 brought the pm_[un]gamma709() library routines and pnm_readpamrown() etc. for manipulating Netpbm images in floating point 0..1 sample values, both linear and gamma-adjusted.

In August 2006, Bryan revised the release strategy so as to have three stabilizing branches (super stable, stable, and advanced) instead of one (stable) in addition to a much more current non-stabilizing branch (development), and distributed all of them via Sourceforge's new Subversion source code control system. He discontinued release via tarball except for the super stable version. The point of this unconventional release mechanism was to reduce the number of bugs in the current releases while minimizing work for the maintainer. The last "advanced" release to have a tarball was 10.34.