Let me share my own stats: I configured only the pregreet test (no rbl, no deep protocol test), and applied selective greylisting (affecting only smtp hosts with a PTR record resembling to a dialup/cable/dsl/... client hostname) after postscreen.
I found that postscreen blocked 15% of the zombies (assuming that postscreen and postgrey together detected all the zombies). Thanks for all of your input on the matter. Kovacs Janos
