There are a bunch in the 644 family larger package
16/32/64/128K Bytes In-System Programmable Flash
with more IO and dual hardware USARTs.
Or did you intend to stay with the smaller package of the 168/328 based boards?
I was playing around with making a small card to mount on top of the 40 pin package herehttp://www.arduino.cc/cgi-bin/yabb2/YaBB.pl?num=1289437587/15
since the leaded surface mount version of the 644 doesn't appear to be currently purchasable.
Build up a little PCB, glue to top of the package, bring 11 wires down to the leads directly underneath some pads. microsystem on a chip :-)
I know others have done similar by installing the parts right on the dip directly, I was going for something that was easier to repeat. Tried selecting decent sized/spaced pads to make hand soldering easier (on me) vs those little microchip sized parts.
Or make a sandwich, have a small breakout board underneath the dip also. That's also shown there.