If the station develops an unexpected rapid loss of pressure, the initial process is to go to a safe location when the alarm is sounded, then begin trying to locate the leak if possible. There is a detailed summary in section 3.1 of this paper.
The threshold for evacuating rather than staying to locate the problem is when the time left before dropping to the minimum safe pressure level (9.5 psi) is less than ten minutes. This seems to be the estimated time required to safely allow everyone to get to their spacecraft seats.
So the threshold for "big enough to cause a catastrophic loss" is probably "a hole that would depressurise the station in ten minutes or less" - anything that causes that big a hole will probably also have caused other problems, but let's leave that aside. Can we work out how large a hole that would be?
To expand my comment on OM's answer, we have two quoted values for estimated depressurisation rates to the 9.5psi threshold:
- 20cm hole = depressurise from 14.7psi to 9.5psi in 50 seconds (not enough time to evacuate)
- 6mm hole = ditto in approximately 14 hours (definitely enough time to evacuate)
If we model this linearly (but see comments on the other answer as to whether this makes sense), then it suggests that a hole of approximately 4-5cm diameter would give a time to minimum safe pressure of 10-20 minutes. This would probably be on the limit of what is survivable for the crew but not the station - the crew would be able to evacuate to a safe location before the pressure dropped too low, but would have to then abandon the station.
A larger hole than this, and the crew might not be able to safely evacuate at all.
A smaller hole than this, and there is a good chance they would have enough time to isolate the leak and seal off the damaged module(s), leaving at least part of the station safely pressurised and operational. I cannot work out how long it would take to go through the full process of finding the leak - probably a couple of hours, but hopefully this question will have an answer.
(Whether you would still want to keep the crew up having had that level of damage is another question, of course.)
I cannot find a clear statement as to whether they would continue putting air into the station to try and maintain a safe level while a rapid depressurisation is ongoing, but I think the answer is probably not. This paper discusses the nitrogen reserves as providing "capability to isolate modules and recover cabin pressure in the remaining volume during a potential “rapid” depress event", but "remaining volume" suggests it would be used to bring the safe area back up to pressure after the leaking section was sealed off, not while it's still losing air.