From my personal experience, unless the book is really oddly formatted, or the scans amazingly clean, it's just not worth OCRing a scanned book if all I plan to do is read it on my Sony myself, especially with all the manual correction that would need to be done after the OCR.
(If I plan to distribute the book to others, that's another ball of wax.)
Scan the book -- if they're available on your OS, run it through something like unpaper or scan tailor, and finally, process the resulting image-based PDF with something like PDFRead or
pdflrf to divide the text into Sony-reader-sized chunks.