1. The measured recidivism rate will depend on time frame. The longer the frame, the higher it is. Most studies use a shorter time frame.
2. Most sex crimes do not get reported. Most studies rely on reported crimes because it is obviously easier to get data for them.
3. Most re-arrests are for non-sexual offenses. But a non-sexual crime when carried out by a pedophile may have a sexual motivation.
That said, in the largest available long term study, the 10 year rearrest for another sex crime is 20%. The actual re-offense rate is certainly much higher.
Making it personal, I was abused as a child. I know two other people who that person abused, and I have strong evidence that there were many more before that. However looking at his legal history, you'd only see a single arrest and conviction for public indecency in the 1960s.
I have seen surveys indicating that the most unusual thing about this example is that he was arrested and convicted at all. That perfectly fits with what I've heard from other people who were abused. Most of the abusers were serial offenders, few were convicted.
Moving on, visit http://virped.org/ to get an idea of how hard it is to be a pedophile who is attempting to never offend. I fully respect their attempts to be better people, but won't be letting my children be around anyone who I suspect is in that group...my reasons should be obvious...
On #3, anyone with experience in how the legal system works will agree that while how often it happens may be hard to quantify, but it certainly does happen. And probably not very rarely.
Got any support for this? I'm finding contradictory information.