Evapotranspiration (ET) was quantified over plant communities within the Big Cypress National Preserve (BCNP) using the eddy covariance method for a period of 3 years from October 2007 to September 2010. Plant communities selected for study included Pine Upland, Wet Prairie, Marsh, Cypress Swamp, and Dwarf Cypress. These plant communities are spatially extensive in southern Florida, and thus, the ET measurements described herein can be applied to other humid subtropical locations such as the Everglades. The 3-year mean annual ET was about 1,000, 1,050, 1,100, 930, and 900 mm (millimeters) at the Dwarf Cypress, Wet Prairie, Cypress Swamp, Pine Upland, and Marsh sites, respectively. Spatial differences in annual ET were considerable due to the recovery of the Marsh site from extensive forest fire and drought conditions. Temporal variability in annual ET was relatively small at sites that were well watered (Dwarf Cypress, Wet Prairie, Cypress Swamp, Pine Upland) over the 3-year study. In other words, locations that were well watered appeared to have similar annual ET rates. Diurnal water-level variability was observed in response to ET and was less at flooded sites than at dry sites. For example, surface-water levels declined about 1.5 mm in response to ET at the flooded Cypress Swamp site during July 18-22, 2008 and declined about 10 mm in response to ET at the flooded Dwarf Cypress site from April 18-27, 2008. Specific yield was computed using ET estimates and diurnal water-level variability measured at the dry Pine Upland site as a check on the accuracy of the eddy covariance method. Water levels repeatedly dropped about 15 mm on average in response to ET at the Pine Upland site from April 27 to May 4, 2008. ET was about 3 mm on each of these days, resulting in a reasonable estimate for specific yield of 0.2 for the sandy soils at the Pine Upland site. Monthly ET estimates exhibited seasonal variation. ET was generally greatest between March to October when solar radiation was relatively large, and least from November to February when solar radiation was small. Monthly ET was greatest in the spring and summer at the Cypress Swamp site, reaching rates as large as 140 mm. The large ET rates at this site coincide with the most active period of cypress growth during late spring and early summer. Cypress trees begin to senesce in late summer reducing transpiration. Net radiation and available energy explained most of the variability in ET observed at all five sites. Mean annual and monthly net radiation varied among the sites in response to cloud cover and the albedo of the land surface and plant community. Net radiation was greatest at the Cypress Swamp site, averaging about 130 W/m² (watts per square meter) during the 3-year study. Net radiation was generally less at the Dwarf Cypress site, averaging about 115 W/m² over 3 years. The Dwarf Cypress site apparently has the largest albedo, which likely is due to the sparse canopy and a highly reflective, calcareous, periphyton-covered land surface. Furthermore, mean annual net radiation was least in the first year of the study, which likely was due to greater cloud cover during a relatively wet year. In contrast, net radiation was greatest in the second year of the study, which likely was due to less cloud cover during a relatively dry year.