Sure. The following quotes are all from statistics compiled by the Bureau of Justice Statistics while under the direction of John Ashcroft, so they hopefully should be relatively free of “liberal bias.” 
Also, I’m going to use “reconviction” rates rather then “rearrest” or “returned to prison” rates–I think that ex-felons are more likely to be arrested for a crime, which I think will skew re-arrest rates against them, and the “returned to prison” rates include parole violations, which generally result from doing things that wouldn’t be illegal if the person wasn’t under parole supervision. The reports themselves report both re-arrest and re-conviction, and include a discussion of differences between the two metrics (they prefer re-arrest rather then re-conviction, because it has a lower burden of proof).
Here we go:
For all prisoners, 46.9% were reconvicted of a new crime within 3 years of their release from prison. Of those, 25.4% went back to prison for that crime. If you divide categorize the released prisoners and enumerate their “reconviction”/“back to prison for new crime” statistics, you get (in decreasing order of decreasing recidivism):
Type of Crime Re-Conviction New Prison Sentence
Larceny/theft 55.70 32.60
Stolen property 57.20 31.80
Motor vehicle theft 54.30 31.30
Burglary 54.20 30.60
Other/unspecified 60.50 28.80
Other property crime 47.60 28.50
Kidnaping 37.80 25.10
Robbery 46.50 25.00
Drug trafficking 44.00 24.80
Other public-order 58.00 24.40
Weapons 46.60 24.30
Drug possession 46.60 23.90
Fraud 42.10 22.80
Assault 44.20 21.00
Other 48.00 20.70
Arson 41.00 20.10
DUI 31.70 16.60
Other violent crime 29.80 12.70
Rape 27.40 12.60
Homicide 20.50 10.80
Other sexual assault 22.30 10.50
Note that lewd acts with a child, etc. are gonna be under “Other sexual assault.” If we’re interested in sex offenses involving children specifically, the percentage of released child molesters re-convicted of a new sex crime within three years is 5.4%.
Why do you hear the “nearly all sex-offenders re-offend” schtick so often? I’d guess it’s because:
[list=a]
[li]It makes sex offenders even more scary and dramatic, and makes for better TV.[/li][li]They often use it to describe out-of-the-blue man-in-a-van-looking-for-his-dog type molesters, who are a minority of child molesters. Most are friends and family of the victim; child molesters who molest unknown children are a minority.[/li][li]One “recidivism” statistic sex offenders do score high in is “likelyhood of commiting the same offense again.” However, this can be a very misleading statistic, because it only counts those who re-offend. If a child molester is released, AND that child molester commits another crime, that crime is more likely to be another child molestation then it is to be, saying, knocking over a 7/11. However, has we have seen, the percentage of child molesters who do reoffend is rather small.[/li][li]The sensationalist statistics are often given as a multiple of another statistic, i.e., “sex offenders are 4 times more likely then other released criminals to commit another sex crime.” It sounds mighty impressive, but if the percentage of released criminals who commit a sex crime upon their release is rather small, then it’s 4 times a rather small number, which is likely to be a rather small number itself.[/li][/list]
These statistics are from the following two reports:
Recidivism of Prisoners Released in 1994
Recidivism of Sex Offenders Released from Prison in 1994