When Amtrak compares its fuel economy with automobiles (see p. 19), it relies on Department of Energy data that presumes 1.6 people per car (see tables 2.13 for cars and 2.14 for Amtrak). But another Department of Energy report points out that cars in intercity travel tend to be more fully loaded — the average turns out to be 2.4 people.
Even if a particular rail proposal did save a little energy in year-to-year operations, studies show that the energy cost of constructing rail lines dwarfs any annual savings. The environmental impact statement for a Portland, Oregon light-rail line found it would take 171 years of annual energy savings to repay the energy cost of construction (they built it anyway).
Public transit buses tend to be some the least energy-efficient vehicles around because agencies tend to buy really big buses (why not? The feds pay for them), and they run around empty much of the time. But private intercity buses are some of the most energy efficient vehicles because the private operators have an incentive to fill them up. A study commissioned by the American Bus Association found that intercity buses use little more than a third as much energy per passenger mile as Amtrak. (The source may seem self-serving, but DOE data estimate intercity buses are even more efficient than that
Often lost in the discussion about rail, is that the energy efficiency can vary quite dramatically based upon the type of rail and it's use. Light rail, commuter rail, high speed rail and freight rail all have very different energy consumption rates.