PDF is very widely supported. Virtually all platforms have a readily available reader. The status of screen reader support is not as clear. The major screen reading programs on Windows and VoiceOver on Mac do a good job reading PDF. On Linux, there are a number of screen reader projects. We have not tested them yet. We have seen mention online of Gnopernicus working well with PDF. KTTS is the text-to-speech system for the KDE desktop. It is designed to work with KDE desktop applications; however, the developer roadmap shows that KPDF is currently not working with KTTS. One of the most promissing Linux accessibility projects is Orca. It is aimed at the GNOME desktop and is currently being developed by Sun Microsystems. For more on Linux and assistive technologies, check out Sun Accessibility Architect Peter Korn's Blog.
Depending on its settings, Acrobat Professional will attempt to run optical character recognition on PDFs that have been originally created by scanning an image of text. We have seen very good results from this process. In terms of accessibility, however, this cannot be counted on. First, Acrobat Reader, the web browser plugin and stand-alone program most users will have on their systems, does not have the OCR functionality. Second, this post-processing does not created tagged PDF.
In many configurations, viewing a PDF while a screen reading program is running will launch the auto-tagging feature within Acrobat and Acrobat Reader. Tagging is the process of marking up the text within a PDF document and applying to it a structure that a screen reading program can understand. This tagging should be done when the PDF is created (more on this later). If it is not done when the PDF is created, Acrobat and Acrobat Reader will try to automatically generate a tag structure for the PDF. Auto-tagging often results in poorly marked up PDF that is non-optimal for screen readers.
The best strategy for accessible PDF is to use PDF Maker to generate a solid tag structure from inside a well-laid out Microsoft Word document. We cover this below.

You can open MS Word's style menu in a pane in the right-hand side of the Word window. Either click on the icon depicting two "A"s to the left of the styles pull down menu or, from the main menu, Format → Styles and Formatting....
Use the Styles and Formatting menu for all text formatting. Doing so will not only save you time and make the appearance of Word documents more consistent, it will also go a long way toward helping create well-structured and accessible PDFs.
One immediately visible benefit comes from using headings in the Styles and Formatting menu. Headings inserted or formatted from that menu become structure when you create a Table of Contents within Word and become bookmarks in the bookmark panel when you output PDF using PDF Maker.

Text alternatives are read by screen reading programs and provide essential information for blind and low-vision users. You can think of text alternatives for images as the equivalent of "ALT" text for images in web pages.

Created this way in MS Word, when exported via PDF Maker, the table of contents in the PDF is hyperlinked, so that clicking any entry in it will take you to that section in the PDF. This and the bookmarks pane help greatly with navigation for screen reader users and persons with mobility impairments.

The caption on a table or graphic can appear either above, below, or to the side of the image. It won't appear in the table of figures, however, until you convert the caption text box to a frame, as illustrated above.



As mentioned earlier in this presentation, there are specific use cases for PDF. If your document is going to be on the web and does not fall into one of those use cases, well structured HTML is generally a more accessible format than PDF.
For a definitive overview of when to use PDF and discussion of its relative accessibility, see Joe Clark's Facts and Opinions About PDF Accessibility. WebAIM also has an excellent discussion and tutorial on PDF Accessibilty.