A team of researchers in Toronto, Canada has recently published a paper, titled, "Child Pornography Offenses Are a Valid Diagnostic Indicator of Pedophilia" in the Journal of Abnormal Psychology (August, 2006, Vol. 115, No. 3, 610-615) averaging their results for this methodology across 685 people (some known child offenders (e.g., known pedophiles), some adult offenders (e.g., standard rapists, etc.), some adult porn addicts, and some convicted of child porn viewing, but otherwise not known to be pedophiles. Their results confirm statistically that a predilection for viewing child porn is closely associated with pedophilia. In fact, study subjects convicted of viewing child porn but not previously convicted of any actual child offenses were almost three times as aroused by child porn photos as actual convicted pedophiles. Adult offenders (e.g., standard rapists), as you might expect, showed arousal far more frequently to adult pictures than to child pictures.
What this study does is to establish that child-porn-seeking behavior is in fact, statistically speaking, powerfully associated with the same sort of sexual arousal patterns (e.g, turned on by children rather than adults) associated with convicted pedophiles. You've heard the phrase, "where there is smoke, there is fire", perhaps? Well, here is mathematical evidence that this association works for child porn and pedophilic behavior too. The presence of child-porn-seeking behavior doesn't prove pedophilia, of course, but when it is present, pedophila is statistically more likely to be the case than not.