It does depend on what you do. Contrary to some of the reports here, the MBP 13" inch is actually quite suitable to certain aspects of heavy lifting that the MBAs can't do as well. Even for the MBP 2009 model, which handles a little worse for games due to the older Nvidia chip. (If the Starcraft 2 demonstrations are to go by, the MBA 13 2GB ram plays the thing at all medium settings better than my 2009 model does with mostly low settings with a couple of medium selections) However, the 2009 model still outperforms them for certain CPU intensive tasks and especially music work, from my tests.
Even these MBAs, in music, may allow you to out basic ideas together, they lack the power to handle large music works (Heck, all the in-store 11 and 13 inch models (2GB ram) start to choke beyond 2 internal instrument tracks in GarageBand), while then MBA 13" late 2009 allows me to push an entire orchestra through the thing without a hitch in Logic Pro Studio. And yes, I do work with 30+ tracks working with multiple layers of realtime channel effects, chewing about a gigabyte of data for 1 song, with all the recorded audio, etc. Now to be fair, there are certain instruments in Garageband (while using Garageband) that unfairly tax the system beyond a larger piece in Logic Pro Studio. My simple test of two tracks of Solo Star (Keyboard Synth) and Upright Bass (while using mac keyboard entry) is actually really taxing on the CPU. (as simple as it sounds to do this test, you'll be surprised that you can actually do a nice small band in Logic Pro Studio well within the power necessary to do these particular 3 instrument tracks together in Garageband. I think it has something to do with the real expensive audio effect layers that default to Solo Star, particularly the echo and reverb ones) But my Macbook Pro still handles it a bit better than the MBAs do. It's at least possible and doable without locking any tracks, while even with locked tracks, the MBAs I tested, were brought to their knees. On the other hand, they would probably do a bit better, with that test, with an actual USB musical keyboard attached since it's less taxing on the system to use. Mac Keyboard entry requires a good deal more resource to avoid input lag (but then again, with a real small unit such as the air, you're likely to want to input your little ideas without the use of a large and bulky external keyboard.
On the other hand, I use my MBP as my main workhorse computer, even though it's also a portable. It's just not as portable as the MBA.
My Garageband tests is actually my own personal benchmark I use, across Macs, to see how well the CPU/HD Access/Memory/Bus speeds work together in harmony. It's actually surprisingly a more taxing test (in a short amount of time) than a lot of folks use for other applications. However, the test doesn't work the Graphics Chip. That test is pretty much specifically for testing everything else.
The only thing I wish I could test, in store, is the 1.6 model MBA 11. While the processor is still not the 1.86 of the 13", I'm curious if the slight performance gain (in this particular test) for the 13" has more to do with the frontside bus difference than the processor. The 13" one has the frontside bus speed to match my Pro.
I'd also wish they had a 4GB in store model to see how that improves working with HD iMovie video in iMovie. Another thing my 13" 2009 model seems to outperform the Air 13" model on. I really suspect it's the amount of ram in the display models.
