Originally Posted by DiapDealer
I've been having decent luck by:
1) Cropping with Acrobat 10 Pro (other croppers only hide headers and footers so they can come back and bite you in the ass later... Acrobat gets rid of them).
2) PDF->HTML with ABBYY FineReader 11 (manually massaging as much as possible before saving HTML).
3) Regex the poop out of that html file in Notepad++ or EditPad-Lite (it's never clear exactly what kind of reg-exps you'll need until you start wading into each file) to get it as clean as possible. Mark/create chapter headers and chapter-points so Sigil can coast later on (it can get cranky, so if I can spare it a bit of grinding... I always try to). Get a basic external stylesheet going and link it in.
3) Open in Sigil. Finalize/tweak CSS and formatting. Split at predefined chapter-markers. Generate/tweak ToC (ncx).... create an HTML ToC if necessary.
4) Proof my eyes bloody.
5) Convert to MOBI with kindlegen.
6) Apply Visine to eyeballs. See if the sun's up yet. If not, crack a beer—if so, put coffee on brew.
As I said to Diap in a Karma, yup--those are pretty much our procedures. Either 1 or 2, and then 3-6, inclusive. That's the only way we have found that is remotely viable (some PDF files really DO export rather loverly in xml, Diap--it's worth a try if you have something relatively simple, FWIW.)