This report compares evapotranspiration estimated with the Bowen-ratio and eddy-correlation methods with evapotranspiration measured by weighing lysimeters for two sparse-canopy sites in eastern Washington. The sites are located in a grassland area (grass lysimeter site) and a sagbrush- covered area (sage lysimeter site) on the Arid Lands Ecology Reserve in Benton County, Washington. Lysimeter data were collected at the sites from August 1990 to November 1994. Bowen-ratio data were collected for varying periods from May 1993 to November 1994. Additional Bowen-ratio data without interchanging air- temperature and vapor-pressure sensors to remove sensor bias (fixed-sensor system) were collected from October 1993 to June 1994. Eddy-correlation data were collected at the grass lysimeter site from March to April 1994, and at the sage lysimeter site from April to May 1994. The comparisons of evapotranspiration determined by the various methods differed considerably, depending on the periods of record being compared and the sites being analyzed. The year 1993 was very wet, with about 50 percent more precipitation than average; 1994 was a very dry year, with only about half the average precipitation. The study showed that on an annual basis, at least in 1994, Bowen-ratio evapotranspiration closely matched lysimeter evapotranspiration. In 1993, Bowen-ratio and lysimeter evapotranspiration comparisons were variable. Evapotranspiration estimated with the Bowen-ratio method averaged 5 percent more than evapotranspiration measured by lysimeters at the grass lysimeter site from October 1993 to November 1994, and 3 percent less than lysimeters at the sage lysimeter site from November 1993 to October 1994. From March 24 to April 5, 1994, at the grass lysimeter site, the Bowen-ratio method estimated 11 percent less, the Bowen-ratio method utilizing the fixed sensor system about 7 percent more, and the eddy-correlation method about 28 percent less evapotranspiration than the lysimeters measured. From May 7 to June 18, 1993, however, the Bowen-ratio method estimated only 54 percent of the evapotranspiration measured by lysimeters at the grass lysimeter site. This large difference possibly may be attributed to Bowen-ratio instrument variability or error, to the density of grasses in the lysimeters being greater than in the surrounding area, or to heating effects on the lysimeters. From September 1 to October 31, 1993, the Bowen-ratio method estimated more than 450 percent more evapotranspiration than was measured by lysimeters at the sage lysimeter site. This difference may have been due to conditions in the lysimeters at the sage lysimeter site that were unrepresentative of natural conditions. The Bowen-ratio instruments measured evapotrans- piration over sagebrush plants outside the lysimeters, which were blooming very heavily, possibly using supplemental ground water or spring water from nearby upslope areas. The sagebrush plants contained by the lysimeters showed very little evapotranspiration, possibly because they were root-bound and had already used all available water. Also, plants in the lysimeters would not have been able to access any supplemental water available to plants outside the confines of the lysimeters. Earlier in 1993, from June 17 to July 12, the Bowen-ratio method estimated only 1 percent less evapotranspiration than determined for the lysimeters at the sage lysimeter site. On the basis of lysimeter measurements from August 1990 to November 1994, cumulative evapotrans- piration ranged from about 97 to 103 percent of the annual precipitation each year. The evapotranspiration measurements made at the grass and sage lysimeter sites, which were based on weight changes in the lysimeters, showed that storage changes became nearly zero each year some time between August and November as average surface soil moisture decreased to about 2 percent and evapotranspiration rates decreased to less than 0.1 millimeter per day.