The precisions of extrasolar planet radius measurements are reaching the point at which meaningful and discriminatory comparisons with theoretical predictions can be made .
However , care must be taken to account for selection effects in the transit surveys that detect the transiting planets for which radius measurements are possible .
Here I identify one such selection effect , such that the number of planets with radius R _ { p } detected in a signal-to-noise ratio limited transit survey is \propto R _ { p } ^ { \alpha } , with \alpha \sim 4 - 6 .
In the presence of a dispersion \sigma in the intrinsic distribution of planet radii , this selection effect translates to bias b in the radii of observed planets .
Detected planets are , on average , larger by a fractional amount b \sim \alpha ( \sigma / \langle R _ { p } \rangle ) ^ { 2 } relative to the mean radius \langle R _ { p } \rangle of the underlying distribution .
I argue that the intrinsic dispersion in planetary radii is likely to be in the range \sigma = ( 0.05 - 0.13 ) R _ { J } , where the lower bound is that expected theoretically solely from the variance in stellar insolation , and the upper bound is the 95 % c.l .
upper limit from the scatter in observed radii .
Assuming an arbitrary but plausible value of \sigma / \langle R _ { p } \rangle \sim 10 \% , and thus b \sim 6 \% , I infer a mean intrinsic radius of close-in massive extrasolar planets of \langle R _ { p } \rangle = ( 1.03 \pm 0.03 ) R _ { J } .
This value reinforces the case for HD209458b having an anomalously large radius , and may be inconsistent with coreless models of irradiated giant planets .