I generally use 2 pass because the results are about as good as they are going to get and more doesn't offer anything better. I've done many an experiment running multiple passes (never more than 6) and the added time that it takes to complete doesn't seem worth it. No matter how many passes one uses in the final analysis there is only a finite amount of available space, and running multiple passes is not going to be that much better and reallocating bitrate.
I have five dual core systems and 1 dual core laptop so I wouldn't loose the use computer time by running multiple passes. This of course is just my opinion, others might see it differently,
I'm from the 3 pass school of thought. If 2 are needed and the 3rd might help, I'm willing to add an extra pass. I do most backups in batch mode, so I don't sit and wait. I have a dual core with decent resources and able to multitask when I want to. So, it's not a mater of time. I just like to insure the best quality possible. Even if the last pass doesn't always improve the output, at least it helps my peace of mind.
I personally use 2 passes with everything.... cause that's the most CCE Basic will do ;)
If I were using an encoder that will allow more than 2 passes I'd probably opt for the extra pass on everything just to make sure. MPEG-2 encoding is pretty easy for any computer made in the last couple of years (actually mine is nearly 5 years old) so I don't see any reason not to make sure I'm doing the best possible job.
