From my experience if there is a light load on the engine, use low rpm. If you need more power, use higher rpm. Overally try to keep the rpm as low as possible without lugging the engine, and without using more than about 3/4 throttle. On fuel injected engines, keeping the throttle light keeps the ecu in closed loop where the A/F ratio stays at 14.7:1. If too much throttle is used, the ecu goes into open loop where it runs extra rich. A basic A/F ratio gage can show when the ecu/engine is in open/closed loop, so you can see when the engine is running outside of it's most efficient zone.
It's good to have more gages so you know what's really happening, instead of guessing. Every car is different, what works for a big engine won't neccesarily apply to a small engine. A supermid, scangage, or even a datalogger are valuable tools.
I don't recomend trying to drive with the engine rpm spinning at peak torque since most engines make peak torque high up in the rpm band. For normal driving on flat roads you shouldn't need maximum torque to move the car. For example, here's a dyno chart similar to mine:
http://www.dsmtuners.com/gallery/sho...searchid=28509
Notice peak torque is at about 3500rpm. Typically I shift gears below 2krpm, sometimes I'll shift at 1500rpm and I'm seeing a nice improvement from when I used to shift around 2,500-3k rpm. Here's the thread where I started shifting at low rpm, the whole thread is good, but the experiment starts at post #23 where I observe fuel use at different rpm with no load,
www.gassavers.org/showthread.php?t=3140
BTW another word about that dyno chart. It shows a certain feature common to turbo cars, that the peak torque occurs about 200 rpm after the turbo reaches full boost. I don't know if that's true of a low boost turbo on an engine that was originally designed without a turbo though.
__________________