I've yet to work on a game project that didn't involve some amount of crunch near the end, so I'm not here to talk about how it's avoidable...even though I do think it can be :)  I am a firm believer that an employee's effectiveness during crunch hours drops dramatically and have seen many times that a 12 hour day can be much less productive than 8 solid hours.

I came across this IGDA Article by Evan Robinson that talks about the ineffectiveness of Crunch and how most industries quit doing this 75 years ago.  When used long term, Crunch actually slows development and creates more bugs than a 40 hour work week.  Working over 21 hours continuously is equivalent to being legally drunk and can result in severe decrease in cognitive ability.

The thing I've learned about myself over many years of working on various projects is that if I don't get some breaks in the crunch...I will begin to get sick.  Once early in my career, I worked my self so ragged that I got Mono and that was no fun at all.  On the Simpsons Game we were working toward our first playable level which I was building.  I was on a really tight deadline and ended up getting so sick that I was taking naps on the couch at work in between builds so I could get some rest.  My illness actually wiped out nearly the entire team...how's that for decrease in productivity?  I even got an award for it!
Now I'm much more aware of what I can handle crunch wise and am careful to take the breaks I need.  While crunch is something that is usually required to some extent on a game project...I highly suggest being conscious of the diminishing returns.  This often means being aware of your own limitations.