Find old websites from the 1990s

A field guide by Khalid Alshaikh · 8 September 2026

GE97 brings together evidence of the early web and ways to explore it. Start with the Early Internet Directory: 75,635 catalogue records, including 74,157 with period descriptions. You can read these pages without booting OS-98. For the desktop experience, open The Old Internet.

A directory entry, a screenshot and a playable website preserve different things. An entry can tell you what a site was called and how people described it, even when no working copy has been found. The catalogue count is not a count of complete, functioning websites.

Where to start

Remember a name or topic? Use directory search for names, subjects and years, or browse the alphabetical and subject lists. The search page does not search the descriptions themselves.
Remember the address? Try it in the Wayback Machine. Also check directory records for an older spelling, host or path. A current website at an old address may belong to somebody else.
Recognize the layout but forget the name? Browse the screenshot gallery. Read image captions: a picture can show another page from the same site rather than the exact address in the record.
Need a preserved site to explore? Visit the 1996 CD-ROM collection, including the 1996 White House exhibit. Copies preserve what was on the disc; outside links and later server functions may be missing.
An old interactive file will not play? See Revived Web for four Director experiences with per-title compatibility notes. In OS-98, choose Revived in Internet Explorer to open them in the running desktop.
Researching the Saudi or Arabic web? Read Al-Qadsiah supporters on the internet (بالعربية), then explore the 2001 site, Kooora material and 2009 site and forum.

How to investigate a missing website

Find the period record. Search a distinctive part of the name; try a broader subject if that fails. Browse descriptions on the linked catalogue pages. A crawler's short extract is different from an editor's review.
Note the evidence. Keep the original address, the displayed year and the source credit. A directory's publication year is evidence of its listing, not necessarily the site's founding date.
Look for a surviving copy. Try archive captures of the listed URL, its home page, and variants with or without www. Use the record's “open in the Old Internet” link to explore through the exhibit. A listing cannot guarantee a capture.
Record what is missing. A screenshot cannot restore a form, a database or a game. If an archive has no result, describe the search you tried; that alone does not prove the site was never saved anywhere.

What makes this collection useful?

The directory combines printed and CD-ROM directory entries, the October 1995 Lycos crawler index, site announcements and Scout reviews. Source credits stay with the records, so you can distinguish a contemporary description from GE97's restoration work. Some entries are duplicate variants, broken addresses or instructions rather than home pages.

The Recoveries log explains the work behind the collections, including address corrections and missing archive evidence. Counts here were checked against the catalogue on 8 September 2026. The descriptions are historical documents; their claims have not all been independently verified.

Can every old website be brought back?

No. A preserved file still needs a compatible player, and missing assets or server code may prevent full restoration. Revived Web distinguishes tested controls from unverified completion and credits Flashpoint's preserved files. Other archives may hold the same material.

How to cite what you find

Cite the named period source and original address alongside the GE97 catalogue page and the date you consulted it. For a screenshot, include its caption; for an interactive exhibit, include its source and compatibility notes. This keeps the historical evidence separate from what the modern player can reproduce.

For another route into the period, explore the 1994 hotlist, vanished sites exhibit and Before the Web.