According to the 2000 data, public water supply, defined as "water withdrawn by public and private water suppliers ... [that] may be delivered to users for domestic, commercial, industrial, or thermoelectric-power purposes", accounts for only 11 percent of total US water use; "domestic use", defined (only) for the purposes of this report, accounts for less than 1 percent of fresh-water withdrawals. Although there is no breakdown in this report, the figures from 1995 would seem to indicate that private residential use makes up around 53% of the public supply category.

So even if we assume that, as Wikipedia claims, "approximately 50-70 percent of American residential water is used for landscaping, most of it to water lawns" (and there is no reason we must, as there is no citation for this assertion), water consumption for landscaping does not account for more than, say, 7-8% of US water use at the very outside.

So where is the water going? According to the USGS 2000 figures, irrigation accounts for 34% (+ 1% livestock) and thermoelectric use was 48% - of which one third was saline.

Where can the most water be saved? Cornell University puts its finger on the issue:

Cornell News: Wasted water resources

ITHACA, N.Y. -- In a world plagued by shortages of water, three facts stand out in an analysis by Cornell University ecologists: Less than 1 percent of water on the planet is fresh water; agriculture in the United States consumes 80 percent of the available fresh water each year; and 60 percent of U.S. water intended for crop irrigation never reaches the crops.

Their report in the October 2004 journal BioScience (Vol. 54, No. 10, "Water Resources: Agricultural and Environmental Issues") names farmers as "the prime target for incentives to conserve water." The report is particularly critical of irrigation practices in the United States, where subsidized "cheap water" offers scant incentive for conservation.

About the 80% figure: the key words are "available water", defined here as "the quantity of renewable water resources available for human use", and "consumption", defined by the same source as "water that is evaporated or incorporated into products and organisms, so that it becomes temporally unavailable to the other users". The USGS figures refer to all water - renewable and otherwise - withdrawn (either consumed or put back in the system). So the 80% figure can probably be considered plausible under consideration of the different counting approaches. (Although the Cornell blurb seems to count the aquifers as "available water", even though their renewability - particularly in the Plains states - is often debatable.)

So - to return to the USGS figures - increasing irrigation efficiency could reduce total US water use by up to 20%. So I'm not too concerned about the lawns.