Mostly; the structural information is in a mix of XHTML and various XML schemas. Conceptually it's more like that weird "HTML archive" format that IE used to do back when, but considerably more complex, and much less terrible thanks to being a W3C standard and not MIME-encoded.
The .epub file itself is just a zip archive. You can open it that way and pull the HTML content out, if you want, although it may be a pain to work with absent some kind of conversion - I think Calibre can linearize an epub into a single HTML file, but I haven't actually needed to do that so can't say for sure.
This is off topic, but I found a recent comment from you just so I could thank you for recommending Free Radical a little while ago. I never even played the game, but the book is excellent on its own.
That's among the reasons why I actually dislike ePub and prefer to convert (which involves pandoc/calibre + some hand-written scripting and hand-editing) epub to FB2 (which I love, it only has some minor imperfections). FB2 is a well-structured straightforward single-file XML designed with data-presentation separation in mind.
I can see the appeal, but that's not an option for me right now, since EPUB is on the input side of the project that's had me studying the relevant standards in the last few days.
That said, it's not the worst standard, if a little bit overcomplicated by trying to be all things to all people with a bunch of features I suspect are almost never actually used in the wild. Having the content already formatted as XHTML is really convenient for what I'm doing with it, too.
The .epub file itself is just a zip archive. You can open it that way and pull the HTML content out, if you want, although it may be a pain to work with absent some kind of conversion - I think Calibre can linearize an epub into a single HTML file, but I haven't actually needed to do that so can't say for sure.