High await numbers

I'm getting very suspicious numbers for the average wait time on my swap partition (hda3). The iostat -x output doesn't look very good wrapped for email, but I'm seeing await over 3000 ms, and svctm over 2000. What's really surprising about this is that the system is using the swap partition, but doesn't seem to be actively swapping - vmstat shows 0 si/so most of the time.

Smart monitoring daemon isn't reporting any problems with the drive, and fsck doesn't show any errors.

I did read somewhere that Redhat's iostat sometimes produces anamolous readings, but the other readings on the drive look reasonable.

Anybody have ideas on how I can pursue this issue further?