HWP vs HWPX: what actually differs
Both extensions come from Hancom Office and both hold the same kind of document, but internally they are built very differently. That difference is the single best predictor of how well your file will convert.
The short version
- .hwp — the older format. A binary OLE2 compound file: one file containing several internal streams, in a layout comparable to the legacy Microsoft .doc format.
- .hwpx — the newer format. A ZIP archive containing XML, comparable in spirit to .docx. Its underlying document markup was standardized, which is a large part of why third-party tools handle it better.
If you have a choice, choose .hwpx. Anything that is not Hancom Office itself will read it more accurately.
How to tell which one you have
The extension is usually enough, but file extensions get renamed and mislabeled more often than you would expect. Two quick checks:
- Copy the file and rename the copy to .zip, then try to open it. If it opens as an archive containing folders and XML files, it is HWPX. If your system reports it as not a valid archive, it is the older binary HWP.
- Look inside the archive. An HWPX contains a `Contents` folder with one XML file per document section — `section0.xml`, `section1.xml`, and so on — plus a `header.xml` holding fonts and styles.
Always work on a copy when doing this. Renaming the original risks confusing whatever program you eventually open it with.
Why HWPX converts more accurately
The practical difference comes down to how much a converter has to guess.
XML inside a ZIP is inspectable. Paragraphs, tables, styles, and section boundaries are named elements a parser can read directly. A binary container requires reverse-engineered knowledge of a structure that was never intended to be read by outside software, so support is inherently more fragile and more likely to lag behind format changes.
There is a second, less obvious advantage. HWPX files saved by Hancom Office typically embed a typesetting cache — the line and page layout the program had already computed. A converter that reads that cache can reproduce the original page layout rather than re-flowing the text and hoping it lands in the same place. This is why a preview of an .hwpx can look nearly identical to the original, while a binary .hwp usually converts to accurate text with approximated layout.
It is worth knowing that this cache is a snapshot, not a guarantee. In our audit of 82 real government .hwpx documents, about 17% produced layout renderings with content clipped outside the page boundary. Accurate text extraction was unaffected — the failure is specific to reproducing visual layout.
Sections, not pages
One structural detail surprises people who try to extract a page range.
These documents are organized into sections, not pages. A section is a run of content sharing page setup — paper size, margins, orientation. Page breaks are a result of laying out a section, not a stored division you can slice at.
This means page extraction on HWP and HWPX is approximate: tools generally operate on section boundaries. Many real documents — press releases and short official notices especially — consist of a single section covering the whole document, so requesting "pages 1-2" from them cannot narrow anything down. When our converter detects that situation it tells you the extraction did not apply, rather than silently returning the whole file as though it had been trimmed.
Which should you ask for?
If you are on the receiving end and can influence what gets sent:
- Ask for PDF if you only need to read and print it. Nothing converts more reliably than a file that needs no conversion.
- Ask for .hwpx if the document must stay editable in Hancom Office but you also need to read it elsewhere.
- Accept .hwp when there is no choice — it still converts to accurate text, Markdown, and PDF. You simply lose exact layout reproduction.