The problem with using overall citywide patronage percentage figures is, as pointed out, that it ignores corridors.

The Manchester Metrolink, for example, carries significantly more passengers than the old rail lines it replaced. (From the article, around 18m per year vs. 5m on the old rail lines). However, a (very anti-rail, pro car) MP tried to rubbish this success by claiming, probably correctly, that Metrolink had only reduced the number of car journeys in Manchester by 1%.

What was not stated was whether Metrolink had meant an increase in the total number of journeys (people now travelling by Metro who didn't previously travel at all), or whether city centre congestion had been reduced (which it probably had).

Agreed.  47% is probably a number from some foaming-at-the-mouth anti-transit site.  All diesel was only 18% of all motor fuels in '95 (the FHWA site seems down this morning, or I'd use more recent numbers).  

To dance around it a bit, the EIA says we use 382.4 million gallons of gasoline/day or ~140 billion gallons a year.  That doesn't include diesel or other motor fuels.

The American Public Transit Association 2005 Factbook says transit uses about 897 million gallons of fossil fuels per year.

That means that transit fuel consumption is .6% of just US gasoline consumption per year.  (Note: it would be much easier to word that if the FHWA site weren't down so I could compare transit's share to all motor fuels instead of just the gasoline portion.)  I expect that Alan's right and .47% is much closer to reality.