Let's see -- Apache, Slash (the /. engine), BugZilla, TWiki, ....

And these are only the ones in the public sector that come immediately to mind.

There are two banks that I personally know of (I've worked contracts with them) with multi-thousand-line Perl back-end systems. I suspect the there are a number of Scientific projects that also fall under the 'large scale' rubric, if only for the volumes of data they are crunching.

Also take a look at this thread from 2006.

Update: Cleaned up the syntax a bit. I gotta stop responding before coffee.