Thanks Mike - that's a tough question, which will take teams of smart people years to answer, and they're working on it. The reason it's hard is that one has to consider all of the elements that go into the "control volume"; that is, the line you draw around your model system where you can measure or calculate all the material and energy that goes in and out. One of the reasons that biofuels have had such a hard time getting started is that these kinds of calculations were difficult to do; just recently a real-world, field scale study of switchgrass farming for fuel was completed (Schmer, M.R., et al., Net energy of cellulosic ethanol from switchgrass. Proceedings of the National Academy of Sciences, 2008. 105(2): p. 464-469.), and they found that you can indeed make more energy than you use to grow it, about 3-5 times as much; the problem is, that energy is not in an easily useable form such as gasoline, so it will take some more engineering to get there.

For solar and wind, I'm not that familiar with what's involved, but producing steel etc., relies on a whole societal infrastructure, so it's much harder to make a control volume - for that matter, how much energy does it take to make a tractor? The easiest thing to do with those types of goods is to put it in terms of money, but of course the price of oil is a major contributor. That said, there are cheaper ways to do things, e.g. recycled steel is a lot cheaper than producing it from iron ore. The main problem with current generation photovoltaic (polycrystalline silicon) cells is that they have a short useful lifetime, although new PV technologies are in development. I believe we currently already have the technology to produce wind and thermal solar power generators that can last essentially indefinitely, with some material and energy input for maintenance, which will greatly reduce their amortized energy cost. And the fuel is free and should last another few billion years ;-)