I have found that PP is more forgiving, both in the amount you can get wrong, and "when" you get them wrong.
PR questions are a bit easier, but if you get anything wrong in the first ten you get a big penalty. Also, they don't consider experimental questions. I got 6 Verbal wrong and 4 Quant wrong (one was completely a PR correcting error) and i ended up with a 670.
On PP1 I got 4 quant wrong and 5 verbal wrong (including the first quant and two out of the first three verbal) and I ended up with a 740.
However, if you use OG before any of the PPs, I do not thing the PP score is very respresentative.
Also of note, Princeton uses questions from its book on the tests, which may inflate scores if you use the book first. I have also heard of them repeating questions on different CAT tests. These companies need to get their acts together (although they won't as long as we're willing to pony up $30 a pop).
Cheers,
Htown