Note that recognition of falling or rising edge interrupts on INT0 requires the presence of an I/O clock, described in "Clock Systems and their Distribution" on page 23.
That has been determined to be incorrect and confirmed by Atmel. I believe Jack Christensen (@JChristensen) discovered the discrepancy and was able to get confirmation from Atmel.
What I cannot remember is if the discrepancy applies to the t85 processor. Knowing Jack's interests I suspect it does.
The discussion took place in the Microcontrollers section.