Many filters from word-processing packages, and also some HTML authoring tools, generate HTML code which is completely contrary to the design goals of the language. What they do is to look at a document almost purely from the point of view of layout, and then mimic that layout in HTML by doing tricks with FONT, BR and (non-breaking spaces). HTML documents are supposed to be structured around items such as paragraphs, headings and lists. Yet some of these documents barely have a paragraph tag in sight!
The problem comes when the content of pages needs to be updated, or given a new layout, or re-cast in XML (which is now to be the new mark-up language). With proper use of HTML, such operations are not difficult, but with a muddle of non-structural tags it's quite a different matter; maintenance tasks become impractical. To correct pages suffering from injudicious use of FONT, try the HTML Tidy program, which will do its best to put things right and generate better and more manageable HTML.
-- World Wide Web Consortium, HyperText Markup Language, Guidelines