The choice of the target document format, into which to convert a given electronic document depends mainly on the contents of this document. Given the impact of the usage pattern of a document, HTML is usually preferrable, but there are reasons which speak for the PDF format in some cases.
With HTML, some things are easy to do, some are difficult to do, and others aren't possible at all, given the current state of the available converters and browsers commonly used, and the definitions on HTML they implement. E. g. documents with lots of formulas end up in including lots of small images into the HTML text, as there's currently no HTML reader in widespread use that supports formulas and mathematical symbols. In addition, available converters as latex2html do not work properly in translating all those formulas into images, so PDF is the way to go here.
Other features such as comments and notes on the margin that the inventors of HTML didn't have in mind but that are common on printed media can be realized in HTML with some effort with tables, frames, and other advanced features. If not, one should investigate if switching over to PDF can solve the problem.
Besides visible aspects, the way in which the document will be used has some impact on what target format to use. While PDF documents are usually downloaded in one piece, documents in the HTML format are downloaded page by page. So if a document is expected to be read from front to back, PDF might be the choice. If it's more likely that only few parts of a document are accessed frequently (like e.g. table of contents and index) and the text is accessed on a on-demand basis, choosing HTML will support this usage pattern in a less network saturating way.