I usually write Lisp in 120 columns. I don't actually put very much on each line, but my code tends to be very deeply nested so it finds its way to the right margin rather quickly. Twenty consecutive close-parens is not unusual for me.
I have the opposite experience. Lisp puts the operator on the far left, and has no penalty for a newline, so I find it much easier to keep lines shorter.
In Algol-based languages, I end up with "something_something [op] something_something" and then if I need to do something else, my next line starts 20 chars from the left edge, to line up with the op. So I have to choose between really long lines, or really short identifiers, or breaking one expression into several using temporary variables -- all of these make me feel dirty.
