I’ve been getting a lot of good results from ABBYY Screenshot Reader. Seems very accurate and it’s easy to use. Display the text to convert on your screen (use PDF reader, pic viewer, whatever it takes). Click the Capture button, drag a rectangle around the text to be OCRed, paste the results where you want. And it’s still free on SOS. It does nothing with formatting, just grabs text. No file browser on input or output, and no preview or correction mode, so Peter B, you may not like it! But it works very well for what it does and in my experience, accuracy is way beyond 98% (at least for English language, I haven’t tried the others it supports). It’s very handy for capturing error message text, parts of web pages that are pictures of text instead of selectable text. Keep in in memory or unload it when you’re done.
I’ve used other OCR programs (from ABBYY and others) and some are very nice — they highlight anything they’re not sure about and let you you fix it before creating the output file. Those programs are all very expensive, but if you have the need to convert longer documents and accuracy is critical, they are well worth it! As for today’s offering, I don’t see what it adds over ABBYY Screen Reader.