Rare event sorting....

I've got a question.  I've got some folks that are needing to sort out small populations.
 .1% or so and smaller.  

How do people prove to themselves that what they've got coming out is what they want when
you're looking at a sorted quantity of 500 cells or so?  At least 
that's what they expect to get in some cases.

I'd like to be sure we're actually doing what we think we're doing.


Thanks,
Chris