The October 2007 issue of National Geographic highlighted this very issue, and it offers a good discussion of some of the same issues addressed in the the more recently published Science findings about the "true" costs of biofuel development. National G did this great analysis of different fuel alternatives and the "energy balance" between the fossil-fuel energy required to make the fuel (input) compared with the energy in the fuel (output). This takes into account fuel for growing and harvesting the fuel as well as the fossil fuel needed to process and refine the biofuel alternative.
Check out some of the energy balance figures (staring p. 44 of the October 2007 issue). All figures are based on a ratio of fossil-fuel input 1.0 : x, so you can simply subtract 1 from the figure to get the net gain in output (input : output). So, for example, for every 1 unit of fossil fuel used, you'll get 1.3 units of corn ethanol.
Corn ethanol: 1.3 (net gain of 0.3)
Cane ethanol (sugar cane): 8.0 (gain of 7.0)
Biodiesel (from canola or soybeans): 2.5 (gain of 1.5)
Cellulosic ethanol: 2.0 to 36.0 (net gains of 1.0 to 35.0, depends on the production method)
Cellulosic ethanol seems to hold the most promise, but as Panoze points out they are still trying to develop large-scale methods for turning cellulosic plant matter into ethanol.
If you have access to it (check your local library!) the October 2007 issue is a good read.