Tom Gardner wrote: I second this recommendation. ABBYY FR is a great product and works fantastically. I don't like many of the auto-modes though, and prefer to select each region of every page manually. This involves clicking a tool, highlighting the area of interest --- and telling abbyy whether it is a graphic or a text box. This process is mind-numbing, but gratifying once it is finished. The OCR is very good, but depending on the type of text, you'll still get sort of 1-5% error. This doesn't sound like alot, but ends up being 4 or 5 corrections per page. Some pages are perfect, and don't require any correction. Proofreading the OCR'd text is absolutely required for a solid finished product. You can almost always tell why abbyy didn't read something --- there is usually a noticeable defect in the original text. There is occasionally "O" for the number zero, and "l" for the number 1 confusion. As you might expect. If you don't tell it (or it didn't figure it out) that a portion of the page is a graphic, it will try to OCR the graphic with worthless although interesting results. :) Abbyy suggests 400dpi minimum, but I usually use 600dpi. I've found that I get pretty good results with 600. There's definitely more errors under 300. I scan to TIFF, of course. :) I've had no luck "training" abbyy to learn the text --- it seems like a never ending challenge, and doesn't improve recognition much that I can tell. Regarding Adobe doing the OCR, their OCR is decent, but no where near abbyy. Adobe OCR is good for "keeping-the-original-scanned-graphic" and make it searchable. But if you entirely recreate the original document errors are simply way too much. The tools for repairing the PDFs are pretty arcane, and Adobe is a just a pain in the butt for this type of stuff. While not perfect, this is an example PDF made from abbyy. http://www.techtravels.org/tech/BrianInstrumentsManuals.pdf Hope this helps. Keith