Be sure to try out at least two different font types in your documents -- a standard body-text font and a monospaced font -- to see how they render on different devices and in different book formats. Sometimes font declarations don't work at all: With the Kindle, for instance, you need to use the HTML <pre> tag in e-books to reliably show text in a monospaced font.
This can be a crucial issue for some books. You need to make sure any illustrations convert correctly depending on the system you're using. Exporting to HTML as an intermediate step helps here, since image references in HTML are honored pretty consistently throughout the conversion process.
Footnotes are typically translated into hyperlinks in e-books, but they also run the risk of disappearing if the conversion process doesn't know how to honor them correctly. This is another reason why exporting to HTML as a first step is a good idea: If footnotes and endnotes render as properly hyperlinked elements in that step, they should remain accessible in the finished product, too.
Some languages -- Japanese, for instance -- use what is called "ruby markup" -- annotations that appear next to the text -- to indicate how certain things are pronounced. HTML supports ruby markup, but that doesn't mean it'll always render correctly in the converted e-book.
There are a number of other curious issues that can arise. For instance, if you have a document where outline headings (which typically indicate chapters) are auto-numbered, the numbering doesn't always survive the conversion process. One document I had automatically added "Chapter __:" to the beginning of each chapter, but once converted into an e-book, the auto-numbering vanished.
Content-creation programs, such as word processors or publishing suites, are only starting to add e-book formats to their lists of possible exports. Most of the time, you'll need to use some kind of standalone application to perform the final conversion.
Some of the tools you might encounter are designed for extremely specific jobs and are not general conversion utilities. Those producing e-books for the Kindle, for instance, need to use Amazon's own e-book tool, called KindleGen, to produce a Kindle-compatible file from HTML or ePub input.
These are only four of the better-known conversion applications; there are a lot more out there. In contrasting their behaviors and capabilities, it's clear we're still a ways from having a single end-to-end suite that fits the majority of users' needs.
Sign up for CIO Asia eNewsletters.