The problems are whether evapotranspiration (ET) from paddy field would be affected by growth stage and by rice varieties, and whether the radiation method and Priestley-Taylor method for estimating ET would be applicable to paddy field.ET was measured with microlysimeter (Tomar and O'Toole, 1980A) at the eight test plots (N1-N8) in which cultivar “Nipponbare” (Japonica rice) was grown under eight different cropping seasons, and at the two plots (K5 and R5) in which cultivars “Milyang 23” (Japonica-Indica cross variety) and “IR-36” (Indica rice) were grown under the ordinary cropping season (see Horie and Sakuratani, 1985, Table 1). Three stainless pans (30cm long, 20cm wide and 60cm deep) were installed in the center part of the each plot: the two with the hills of rice plant and the other without crop were used as evapotranspirometer and evaporimeter, respectively. ET and transpiration (T) from each plot were determined, respectively, by averaging the values of ET from the two pans, and by subtracting the evaporation (E) from the ET. Meteorological data were also collected.In each cropping season of “Nipponbare” rice, the E at the early growth stage in which the leaf area index (LAI) was small, was comparable to the ET at the maximum LAI stage (Fig. 2A). Thus, there was no appreciable difference in ET between the two stages. This also caused the noticeable increase of total E for entire growing periods in the plots of the earlier cropping seasons such as N1 and N2 in which the initial growth was delayed by lower air temperature (Table 1). The mean value of daily ET ranged between 3.5 and 4.2mm day-1 for the eight cropping seasons, averaging 4.0mm day-1.The ratio of ET to reference ET (ETR) calculated from modified Penman method (Doorenbos and Pruitt, 1977) was fairly constant throughout the main growth period except mid-September to October and averaged 1.05 for all the cropping season (Fig. 2B). The overestimate in the ET in the fall seemed to be owing to the enrichment of ET by advective energy and to the underestimate in the ETR. On the other hand, the ET/ETR for “Milyang 23” and “IR-36” tended to increase slightly around the heading stage (Fig. 4). These varieties also tended to show higher T than “Nipponbare” under the LAI above 2.0 and high radiation conditions (Fig. 5). These results suggested that crop factor such as canopy resistance is somewhat important in controlling ET of Indica type rice, whereas not so important for Japonica rice.The variation of the daily value of a in the radiation method (Eq. 7) was relatively small throughout the main growth period (Fig. 6). The linear regression technique for all the cropping seasons yielded a=0.88, and five-day ET was estimated with high correlation coefficient (r=0.912).The daily value of α in Priestley-Taylor method (Eq. 8) greatly fluctuated over the growing season (Fig. 6). Apparently, this variation of α was attributed to that of daily net radiation caused by long wave radiation exchange at nighttime. Priestley-Taylor method seemed to be less effective in estimating ET of paddy field in Japan when daily values of net radiation are used.