HTML

What HTML is

HTML, short for HyperText Markup Language, is the language web pages are written in. It is a markup language: a set of labels, called tags, wrapped around ordinary text to tell a web browser how the page is structured, such as which line is a heading, which block is a paragraph and which words are a link.¹

“Hypertext” means text containing links that connect pages to one another, within one site or between sites.² Those links are what join separate documents into the World Wide Web.

HTML was one of three technologies Tim Berners-Lee had written by October 1990. The other two were the URI (a unique address for each resource, usually called a URL) and HTTP (the protocol that fetches linked resources across the web).³

HTML describes structure and meaning. A page’s appearance is normally handled by CSS, and its behavior by JavaScript.²

It is close to universal. W3Techs detected HTML on 97.8% of the websites whose markup language it could identify, and XHTML (an XML-based variant explained below) on 2.9%, as of October 22, 2026. A site can use both, so the figures overlap.⁴

Elements, tags and attributes

An HTML page is made of elements. A typical element has three parts: an opening tag such as <p>, the content, and a closing tag with a forward slash, such as </p>. Written out, a paragraph looks like <p>My cat is very grumpy</p>.¹

Attributes add extra information inside the opening tag. An image element names its file with src and gives a text description with alt, for example <img src=”logo.png” alt=”Company logo”>.¹

Some attributes are simply on or off. Writing disabled on a form field switches it off; leaving the word out leaves the button working. MDN advises always putting attribute values in quotes, because an unquoted value containing spaces is read as several separate attributes.¹

Spacing in the source file does not carry into the page. In almost every element, the browser collapses each run of spaces and line breaks into a single space, so indentation exists only to make the code easier for people to read.¹

Some elements, called void elements, cannot contain anything and so have no closing tag. The line break <br> is one.¹

Elements sit inside other elements, which is called nesting. They must close in the reverse order they opened; when tags overlap, the browser has to guess what the author meant.¹

Every page follows the same skeleton. It starts with <!doctype html>, which MDN calls “a historical artifact that needs to be included for everything else to work correctly.” The <html> element wraps everything. Inside it, <head> holds information visitors do not see on the page, such as the character encoding and the <title> shown in the browser tab, and <body> holds everything that is displayed.¹

The table below shows a few common elements and the earliest source cited in this entry that describes each one.

Origins at CERN (1989 to 1993)

Tim Berners-Lee’s proposal “Information Management: A Proposal” circulated at CERN, the European particle physics laboratory near Geneva, in March 1989.¹¹ CERN dates the invention of the web to that year.¹²

In October 1990 he started writing a browser and editor called WorldWideWeb on a NeXT computer. In November 1990 the first web server, nxoc01.cern.ch, and the first web page went online.¹¹

That first program could edit pages as well as display them. Berners-Lee has described it as a graphical point-and-click browser in which users could also edit pages and create links.¹³

HTML was modeled on SGML (Standard Generalized Markup Language), an international standard, ISO 8879:1986, for describing the structure of documents.⁷ According to Dave Raggett’s history of the language, tags such as P, H1 to H6, UL and LI came from SGML practice, while the anchor element with its HREF link was Berners-Lee’s own addition.¹⁴

An early CERN page titled “Tags used in HTML” survives in the copy of the project’s pages that W3C preserved as they stood on November 3, 1992. It lists TITLE, A (the anchor), P, headings H1 to H6, ADDRESS, ISINDEX for searchable documents, and several list types including UL, MENU and DIR.⁵

The same page notes that HTML files were then sent without SGML’s standard framing tags, and that the heading tags followed a CERN guide to SGML.⁵

Discussion moved into the open in September 1991, when the www-talk mailing list was set up.¹⁴ On April 30, 1993, CERN put the World Wide Web software in the public domain.¹²

A major early addition came from outside CERN. In February 1993 the National Center for Supercomputing Applications (NCSA) released the first alpha version of Marc Andreessen’s Mosaic browser for X Window computers.¹¹

On February 25, 1993, Andreessen proposed an IMG tag on www-talk that “names a bitmap or pixmap file for the browser to attempt to pull over the network and interpret as an image.”⁶ Raggett records that others were unenthusiastic, but NCSA Mosaic shipped the design anyway.¹⁴

Raggett also records that Mosaic 1, released for Sun workstations in April 1993, displayed images, nested lists and fill-out forms, and that Lou Montulli released Lynx 2.0a, a text-only browser, in March 1993.¹⁴ By September 1993 NCSA had working versions of Mosaic for X, Windows PCs and the Macintosh.¹¹

In June 1993 Berners-Lee and Daniel Connolly published an Internet-Draft, a working document of the IETF (Internet Engineering Task Force), that described HTML and its tags.¹⁵

From drafts to standards: HTML 2.0 to 4.01

After visiting Berners-Lee at CERN in 1992, Dave Raggett drafted a richer version called HTML+.¹⁴ It fed into an HTML 3 draft he published in March 1995, with figures, a STYLE tag and a CLASS attribute. It was judged too large and too full of new proposals, browsers each implemented a different subset of it, and the effort was eventually abandoned.¹⁴

Meanwhile the World Wide Web Consortium (W3C) was founded on October 1, 1994.¹¹ The W3C’s own history later recorded that HTML 3.0 was dropped in favor of HTML 3.2.¹⁶

The first formal standard came from the IETF. RFC 1866, “Hypertext Markup Language 2.0,” was published in November 1995 by Berners-Lee and Connolly. An RFC (Request for Comments) is the IETF’s numbered document series. RFC 1866 defined HTML as an application of SGML, added forms, and said it “roughly corresponds to the capabilities of HTML in common use prior to June 1994.”⁷

The groundwork was a cleanup. Connolly and colleagues collected the tags in wide use, and Connolly wrote a Document Type Definition (DTD) for them: a formal SGML grammar stating which elements may appear where.¹⁴

The IETF’s HTML working group was wound up in December 1995. Raggett dates the W3C’s HTML Editorial Review Board, chaired by Connolly, to February 1996,¹⁴ while Bert Bos’s history of CSS places its creation at the end of 1995.¹⁷

HTML 3.2 became a W3C Recommendation on January 14, 1997. Written by Raggett and developed in early 1996 with vendors including IBM, Microsoft, Netscape and Sun Microsystems, it added tables, applets (small embedded programs) and text that flows around images.⁸

HTML 3.2 also standardized the FONT element for changing text size and color, and reserved the STYLE and SCRIPT elements as placeholders for style sheets and client-side scripts to come in later versions.⁸

HTML 4.0 followed on December 18, 1997, and HTML 4.01 on December 24, 1999, both edited by Raggett, Arnaud Le Hors and Ian Jacobs.¹⁸¹⁹ Work on HTML 4, under the code name “Cougar,” had begun in December 1996.¹⁴ HTML 4 brought style sheets, scripting, better printing, features for people with disabilities and support for more of the world’s languages.¹⁹

HTML 4.01 marked FONT, BASEFONT and CENTER as deprecated, meaning still recognized but on the way out. The specification discouraged the remaining font and alignment elements as well, recommending style sheets instead.²⁰

In June 2000, RFC 2854 registered the text/html media type that servers use to label HTML files, listed .html and .htm as the usual file extensions, and retired RFC 1866.²¹

The browser-war extensions

Browser makers also competed by adding features of their own. Andreessen and colleagues left NCSA in March 1994 to found Mosaic Communications, which became Netscape.¹¹ Raggett notes that Netscape added its own tags without discussing them publicly first.¹⁴

Netscape Navigator’s first beta was announced on October 13, 1994, and it supported new presentational tags such as CENTER.²² In 1995, academics objected to presentational features such as BGCOLOR (page background color) and FONT FACE (choosing a typeface).¹⁴

Microsoft announced Internet Explorer 1.0 in August 1995 and added its own extensions, including MARQUEE for scrolling text. Netscape proposed frames, which split a window into separately loaded panes, in September 1995.¹⁴

Vendors also met to settle differences. In November 1995 browser makers gathered near Chicago to work on standardizing HTML.¹⁴ In April 1996 the W3C review board chose the OBJECT element as a general way to embed content, and Raggett wrote a working draft on scripting derived from Netscape’s JavaScript extensions.¹⁴

The same board left BLINK and MARQUEE out of the standard.¹⁴ This period is covered in more depth in the browser wars entry.

Håkon Wium Lie, who proposed CSS in 1994, later argued that such presentational tags undermined device independence and accessibility, and made content harder to reuse.²²

The XHTML years (1998 to 2009)

In 1998 the W3C membership voted to stop evolving HTML and to build an XML-based successor called XHTML instead.¹⁶ XML (Extensible Markup Language) is a stricter, general-purpose markup format that software can process with fewer guesses.

XHTML 1.0 became a W3C Recommendation on January 26, 2000. It reformulated HTML 4 as an application of XML 1.0, with three DTDs matching the three flavors of HTML 4.²³

The practical difference was strictness. XHTML documents had to be well formed: element and attribute names in lowercase, end tags required, attribute values always quoted, and empty elements closed with a slash, as in <br />.²⁴

Other work continued alongside. The programming interfaces browsers offered to scripts were standardized as DOM Level 1 in 1998 and DOM Level 2 between 2000 and 2003. In 2003 the W3C’s XForms, presented as the next generation of web forms, renewed interest in evolving HTML itself.¹⁶

Some browser makers wanted a different direction. In June 2004, at a W3C workshop in San Jose, California, on web applications and compound documents,²⁵ Mozilla and Opera proposed evolving HTML instead. The W3C’s history records that “the proposal was rejected on the grounds that the proposal conflicted with the previously chosen direction.”¹⁶

The W3C eventually changed course. It let the XHTML 2 Working Group’s charter expire at the end of 2009 without renewal and kept its HTML Working Group going.²⁶

In March 2018 the W3C formally marked HTML 3.2, HTML 4.01 and XHTML 1.0 as superseded, telling implementers to follow the latest HTML specification instead.⁸¹⁹²³

WHATWG and HTML5 (2004 to 2014)

After the 2004 workshop, individuals from Apple, the Mozilla Foundation and Opera Software founded the WHATWG (Web Hypertext Application Technology Working Group).²⁷ Its FAQ cites concerns about the W3C’s direction with XHTML, a lack of interest in HTML, and what the founders saw as disregard for the needs of working web developers.²⁷

On June 4, 2004, Ian Hickson announced the group’s open mailing list, describing the group as a loose, unofficial and open collaboration of browser makers and other interested parties. Its first documents were Web Forms 2.0, Web Apps 1.0 and Web Controls 1.0.²⁸

The W3C signaled renewed interest in 2006, and in 2007 it formed a working group to develop the HTML5 specification together with the WHATWG.¹⁶

The two groups then diverged on how to publish. On January 19, 2011, Hickson announced that the WHATWG specification would be called simply “HTML” and treated as a living standard.²⁹ The W3C still wanted a fixed, numbered HTML5 Recommendation.³⁰ In mid-2012 it brought in a new editing team to finish HTML5 and prepare a draft of the next version.¹⁶

The W3C published HTML5 as a Recommendation on October 28, 2014, calling it “the 5th major revision of the core language of the World Wide Web.” Its editors included Hickson for the WHATWG and editors from W3C, Microsoft, Apple and The Paciello Group.⁹

HTML5 added elements for applications and media. Among them were <video> and <audio>, <canvas>, and structural elements such as <article>, <section> and <nav>.⁹

The <video> element embeds a media player, can offer several file formats for the browser to choose from, and accepts caption tracks. MDN lists it as available across browsers since July 2015.³¹

The <canvas> element is a drawing surface that JavaScript controls through the Canvas API or WebGL, used to draw graphics and animations.³²

These features reduced the need for browser plug-ins. When Adobe announced in July 2017 that it would stop updating and distributing Flash Player at the end of 2020, it pointed to the maturing of open standards such as HTML5, WebGL and WebAssembly.³³

HTML5 also changed what kind of standard HTML is. Earlier versions were defined as SGML applications; the new specification returned to a non-SGML basis because that link had caused confusion between validators and browsers.³⁴

In its place, the standard defines parsing rules for every HTML document, whether or not it is written correctly, and gives parse errors well-defined handling. Browsers must use those rules to build a page’s DOM tree from any file served as text/html.³⁴

The short HTML5 doctype became the norm. HTTP Archive’s 2024 Web Almanac found <!DOCTYPE html> on 92.8% of mobile pages and 91.7% of desktop pages.³⁵

The Living Standard and who maintains HTML now

A WHATWG Living Standard has no version numbers.²⁹ It is updated continuously as feedback arrives, and new features are added over time.²⁷ The HTML Living Standard showed a “last updated” date of October 20, 2026, when this entry was checked.³⁰

In 2017, Apple, Google, Microsoft and Mozilla set up a governance structure and an intellectual property policy for the WHATWG.²⁷ The Steering Group agreement, dated November 30, 2017, names those four companies as its initial members.³⁶

The split with the W3C ended with a memorandum of understanding signed on May 28, 2019. It states that “HTML and DOM shall be developed principally in the WHATWG.”³⁷ The DOM (Document Object Model) is the in-memory representation of a page that scripts read and change.³⁸

Under the agreement, a W3C HTML Working Group raises issues and can endorse WHATWG “Review Drafts” as W3C Recommendations, and the W3C stopped its separate HTML 5.3 work.³⁷ The W3C endorsed its first such HTML snapshot on January 28, 2021.³⁹

The standard keeps growing. The <dialog> element, for modal and non-modal dialog boxes, has worked across major browsers since March 2022.¹⁰ The popover attribute, which turns an element into a pop-up panel, has worked in the latest browser versions since April 2024.⁴⁰

Browser makers also coordinate through the yearly Interop project. Interop 2026, run by Apple, Google, Igalia, Microsoft and Mozilla, lists dialogs and popovers among its focus areas.⁴¹

Accessibility basics

HTML carries much of a page’s accessibility, meaning how well people with disabilities can use it. Screen readers, programs that read a page aloud or output it as braille, rely on the markup to know what each part is.⁴²⁴³

Images need text alternatives. W3C’s Web Accessibility Initiative (WAI) states that “images must have text alternatives that describe the information or function represented by them.” Purely decorative images should have an empty alt attribute, alt=””, so assistive technology skips them.⁴²

Structure matters too. Headings and landmark regions, such as a page’s main content area, let screen reader users go straight to the main content and to the sections that matter to them.⁴³

The page language should be declared with the lang attribute on the <html> element. HTTP Archive found no lang attribute on 13% of pages in 2024.³⁵

The W3C’s Web Content Accessibility Guidelines (WCAG) are the W3C’s standard for accessible web content. WCAG 2.2 was published on October 5, 2023.⁴⁴

Most sites still fall short. WebAIM’s 2026 scan of the top 1,000,000 home pages, run in February 2026, detected WCAG 2 failures on 95.9% of them, with an average of 56.1 detected errors per page.⁴⁵

Several of the most common failures are HTML problems:⁴⁵

• Missing alternative text for images: 53.1% of home pages.
• Missing labels on form fields: 51%.
• Empty links, with no text a screen reader can announce: 46.3%.
• Empty buttons: 30.6%.
• Missing document language: 13.5%.

Generic containers still dominate markup. The 2024 Web Almanac found that <div>, an element with no built-in meaning, made up about 28.7% of all elements on mobile pages.³⁵

How website builders generate HTML

Most site owners never type HTML. Berners-Lee himself has argued that easy hypertext editors, which let people make links with the mouse, would help the web.¹³

On most sites, a website builder or content management system writes the HTML. Two examples show how differently platforms store it and how much of it you can take with you.

WordPress‘s block editor keeps each post as a tree of blocks while you edit. On save, it turns that tree into HTML in the post_content database field, marking where each block starts and ends with HTML comments such as <!– wp:image –>. Each block’s settings are stored as JSON inside the comment, out of sight of visitors.⁴⁶

Because the stored content is HTML, older themes and tools can read it without conversion.⁴⁶

Webflow takes a different approach. On a Workspace plan it can export a site as a ZIP file of HTML pages plus CSS, JavaScript and image folders. The export leaves out CMS collection content, form processing, site search, ecommerce data and user accounts.⁴⁷

The size of the generated HTML varies widely. The 2024 Web Almanac measured a median of 594 elements and 32 distinct element types per mobile page, with a median HTML transfer size of 32 kB. The median element count was down from 653 in 2022.³⁵

After <div>, the most common elements were links (<a>, 12.6% of elements), <span> (11.2%), list items (7.7%) and <script> (3.9%).³⁵

Anyone can test a page’s markup against the standard with the W3C’s Nu Html Checker. It is meant to catch unintended mistakes so they can be fixed, and describes itself as “intended solely as a checker, not as a pass/fail certification mechanism.”⁴⁸

Timeline

1. October 20, 2026HTML Living Standard shows its latest update date at the time of writing³⁰
2. February 12, 2026Interop 2026 announced, with dialogs and popovers among its focus areas⁴¹
3. April 2024The popover attribute works across the latest major browsers⁴⁰
4. January 28, 2021W3C endorses its first WHATWG HTML snapshot as a Recommendation³⁹
5. May 28, 2019W3C and WHATWG sign agreement: HTML developed principally at the WHATWG³⁷
6. November 30, 2017WHATWG Steering Group agreement signed by Apple, Google, Microsoft and Mozilla³⁶
7. October 28, 2014HTML5 becomes a W3C Recommendation⁹
8. January 19, 2011WHATWG renames HTML5 to "HTML" and calls it a living standard²⁹
9. 2009W3C lets the XHTML 2 Working Group charter expire at year end²⁶
10. June 4, 2004Ian Hickson announces the WHATWG's open mailing list²⁸
11. January 26, 2000XHTML 1.0 becomes a W3C Recommendation²³
12. December 24, 1999HTML 4.01 becomes a W3C Recommendation¹⁹
13. December 18, 1997HTML 4.0 becomes a W3C Recommendation¹⁸
14. January 14, 1997HTML 3.2 becomes a W3C Recommendation, adding tables⁸
15. November 1995RFC 1866 publishes HTML 2.0, the first HTML standard⁷
16. October 1, 1994World Wide Web Consortium (W3C) founded¹¹
17. June 1993Berners-Lee and Connolly publish an HTML Internet-Draft at the IETF¹⁵
18. February 25, 1993Marc Andreessen proposes the IMG tag on www-talk⁶
19. November 3, 1992Date of the preserved CERN page "Tags used in HTML"⁵
20. November 1990First web server and first web page go online at CERN¹¹

What it means for your website

See also

• CSS
• JavaScript
• WHATWG (coming soon)
• World Wide Web Consortium (W3C) (coming soon)
• Browser wars
• WCAG
• How does a website work
• Web accessibility

References

1. Basic HTML syntax. MDN Web Docs (Mozilla). accessed October 22, 2026.
2. HTML: HyperText Markup Language. MDN Web Docs (Mozilla). accessed October 22, 2026.
3. History of the Web. World Wide Web Foundation. accessed October 22, 2026.
4. Usage statistics of markup languages for websites. W3Techs. 2026-10-22. accessed October 22, 2026.
5. Tags used in HTML (WWW project pages as of November 3, 1992). CERN, preserved by W3C. 1992. accessed October 22, 2026.
6. proposed new tag: IMG (www-talk mailing list). Marc Andreessen, archived by the World Wide Web History Project. 1993-02-25. accessed October 22, 2026.
7. RFC 1866: Hypertext Markup Language 2.0. IETF (T. Berners-Lee, D. Connolly). 1995-11. accessed October 22, 2026.
8. HTML 3.2 Reference Specification. W3C (Dave Raggett). 1997-01-14. accessed October 22, 2026.
9. HTML5: A vocabulary and associated APIs for HTML and XHTML (W3C Recommendation). W3C. 2014-10-28. accessed October 22, 2026.
10. <dialog>: The Dialog element. MDN Web Docs (Mozilla). accessed October 22, 2026.
11. A Little History of the World Wide Web. W3C. accessed October 22, 2026.
12. The birth of the Web. CERN. accessed October 22, 2026.
13. Frequently asked questions by the press: Tim Berners-Lee. W3C. accessed October 22, 2026.
14. A history of HTML (chapter 2 of Raggett on HTML 4). W3C (Dave Raggett, Jenny Lam, Ian Alexander, Michael Kmiec). 1998. accessed October 22, 2026.
15. Hypertext Markup Language (HTML): A Representation of Textual Information and MetaInformation for Retrieval and Interchange (Internet-Draft). IETF (T. Berners-Lee, D. Connolly), archived by W3C. 1993-06. accessed October 22, 2026.
16. HTML5 (W3C Recommendation), section 1.4: History. W3C. 2014-10-28. accessed October 22, 2026.
17. A brief history of CSS until 2016. W3C (Bert Bos). 2016-12-17. accessed October 22, 2026.
18. HTML 4.0 Specification (W3C Recommendation). W3C. 1997-12-18. accessed October 22, 2026.
19. HTML 4.01 Specification (W3C Recommendation). W3C. 1999-12-24. accessed October 22, 2026.
20. HTML 4.01 Specification, 15: Alignment, font styles, and horizontal rules. W3C. 1999-12-24. accessed October 22, 2026.
21. RFC 2854: The 'text/html' Media Type. IETF (D. Connolly, L. Masinter). 2000-06. accessed October 22, 2026.
22. Cascading Style Sheets (PhD thesis). Håkon Wium Lie, University of Oslo. 2005. accessed October 22, 2026.
23. XHTML 1.0: The Extensible HyperText Markup Language (Second Edition). W3C. 2000-01-26. accessed October 22, 2026.
24. XHTML 1.0, section 4: Differences with HTML 4. W3C. 2002-08-01. accessed October 22, 2026.
25. W3C Workshop on Web Applications and Compound Documents. W3C. 2004-06. accessed October 22, 2026.
26. FAQ about the future of XHTML. W3C. 2009. accessed October 22, 2026.
27. FAQ. WHATWG. accessed October 22, 2026.
28. WHAT open mailing list announcement. WHATWG (Ian Hickson). 2004-06-04. accessed October 22, 2026.
29. HTML is the new HTML5. The WHATWG Blog (Ian Hickson). 2011-01-19. accessed October 22, 2026.
30. HTML Living Standard, section 1: Introduction. WHATWG. 2026-10-20. accessed October 22, 2026.
31. <video>: The Video Embed element. MDN Web Docs (Mozilla). accessed October 22, 2026.
32. <canvas>: The Graphics Canvas element. MDN Web Docs (Mozilla). accessed October 22, 2026.
33. Flash & The Future of Interactive Content. Adobe. 2017-07-25. accessed October 22, 2026.
34. HTML Living Standard, section 13.2: Parsing HTML documents. WHATWG. 2026-10-20. accessed October 22, 2026.
35. Web Almanac 2024: Markup. HTTP Archive (Estela Franco). 2024-11-11. accessed October 22, 2026.
36. WHATWG Steering Group Agreement. WHATWG. 2017-11-30. accessed October 22, 2026.
37. Memorandum of Understanding Between W3C and WHATWG. W3C and WHATWG. 2019-05-28. accessed October 22, 2026.
38. Introduction to the DOM. MDN Web Docs (Mozilla). accessed October 22, 2026.
39. WHATWG Review Drafts of HTML and DOM endorsed as W3C Recommendations. W3C (Philippe le Hegaret). 2021-01-28. accessed October 22, 2026.
40. popover (global attribute). MDN Web Docs (Mozilla). accessed October 22, 2026.
41. Interop 2026: Continuing to improve the web for developers. web.dev (Google, Rachel Andrew). 2026-02-12. accessed October 22, 2026.
42. Images Tutorial. W3C Web Accessibility Initiative. accessed October 22, 2026.
43. Page Structure Tutorial. W3C Web Accessibility Initiative. accessed October 22, 2026.
44. WCAG 2 Overview. W3C Web Accessibility Initiative. 2026-09-17. accessed October 22, 2026.
45. The WebAIM Million: The 2026 report on the accessibility of the top 1,000,000 home pages. WebAIM. 2026. accessed October 22, 2026.
46. Data flow and data format (Block Editor Handbook). WordPress.org. accessed October 22, 2026.
47. How do I export my Webflow site code?. Webflow Help Center. accessed October 22, 2026.
48. About the Nu Html Checker. W3C. accessed October 22, 2026.

Web Building Encyclopedia. First published October 22, 2026. Last reviewed October 22, 2026. Every fact on this page is linked to a source we opened. Spotted an error? Write to [email protected].