WalkScore.com will tell you how walkable a neighborhood is. Put in an address you're considering living at, and it'll tell you which grocery stores, restaurants, coffee shops, bars, movie theaters, schools, parks, libraries, bookstores, "fitness", drugs stores, hardware stores, and clothing and music stores are within walking distance. (They don't seem to have a firm cutoff for "walking distance"; I suspect something counts more towards walking distance if it's closer.) It uses this information to generate a score out of 100 telling how walkable the neighborhood is; it seems to correlate pretty well with my subjective impressions of places I've lived or am otherwise familiar with, although most places I'm familiar with are towards either end of their scale and I'd like some more data from the middle.
(Incidentally, it is possible to get a score of 100, though I'd doubted it for a while; Rittenhouse Square in Philadelphia is an example.)
The biggest problem is that they use "as the crow flies" distances; this was noticeable for one place I've lived which happened to be near a river (the Charles) but midway between two of the bridges across it (the Harvard and Longfellow). In most cases this doesn't seem that serious, because of a fact about development patterns -- the places with "inefficient" street patterns (where as-the-crow-flies distances tend to be the worst understimates) are places with low population density anyway.
There's an interesting picture illustrating the efficiency of grids of streets, showing all the places within a one-mile walk (via streets) from a point in downtown Seattle and similarly from a point in the Seattle suburb of Bellevue. The area (in the plane) filled in in the case of the grid is much larger. It makes me wonder -- how much gasoline could we save by designing in such a way that there weren't so many damn cul-de-sacs?
I recently came across Michael Bluejay's claim that walking is less efficient than driving, in terms of fuel consumption, if you eat like a typical American (i. e. lots of meat). This is because walkers need more food than drivers. It looks like a vegetarian walking is more fuel-efficient than a typical car, and that cyclists, whether meat-eating or vegetarian, are more fuel-efficient than the typical car.
However, as Bluejay points out, this is only true on the level of the individual person who has to make a trip to a certain place. On the level of a whole society, if we develop our communities in such a way that they're friendly to walking, people will travel less miles -- because our communities will be more compact, so one can get to the same number of different places with less travel. In fact, car-unfriendly communities are necessarily more compact, simply because less space is taken up by roads and parking lots! I suspect there's a "critical density" of some sort below which a vicious cycle starts: enough people have cars that businesses feel they need to offer free parking, which lowers the population density, which means even more people get cars, which means more free parking, and so on.