Models of water use for irrigation scheduling require the validation of the evapotranspiration (ET) submodel. Evapotranspiration was measured for a reference crop, savannah grass (Axonopus compressus) and maize (Zea mays L.) grown over three seasons at The University of the West Indies Field Station in Trinidad. A non-weighing lysimeter was utilized in the study. Weather data from the University Field Station were used to compute monthly ET values of the reference crop (ETo) using four prediction equations: Penman-Monteith and the FAO-modified versions of Blaney-Criddle, Radiation, and Penman methods. Comparisons between the predicted ETo from these models and the measured ETo from the grass lysimeter revealed that the Penman-Monteith method was the best followed by the FAO-modified Blaney-Criddle, Radiation, and Penman methods in that order. Crop coefficient, Kc, was computed for the three seasons of maize by dividing the measured crop ET by the corresponding measured ETo. Average Kc values for the three seasons ranged from 0.91 to 1.21 and these were generally higher than the range of values (0.72 to 1.07) for the published Kc values based on the Penman method. Evapotranspiration and Kcs for maize varied considerably with season.