>Like Rick said, use an external component. I've tried TextControl and TextDynamic. Back then (8 and 13 years ago) both were very capable and could be installed easily, the startup code with registration variables setting wasn't too complicated and both cost around 500€ for an activeX control. Nowadays both have .net versions. I'm still using TextControl for mass conversion of docx to pdf in several clients. Mind you, any floating frames in the .docx file will throw off the calculation of positions and there may be empty pages in the .pdf; everything else renders fine, small memory consumption and no leaks.
>The method of using Word itself may be OK for one document or two, but then oWord.quit() or there will be problems, sooner or later. When I tried to automate docx-to-pdf i Word itself, it was consuming about 20M per document (of perhaps 200K size in the average, containing perhaps 2K of text), which it never returned. I had to shut and relaunch Word at least on every 200 files or it would start crashing.

Automation from the Web is to be avoided. I will look at TextControl.

Thanks