Methane emission rates from wetland rice fields were determined in Los Ba¿os (Philippines) using an automatic system that allows continuous measurements over time. Methane emission was monitored in an irrigated Aquandic Epiaqualf planted to rice cultivar IR72. Urea fertilizer was applied using four modes: (1) broadcast 10 days after transplanting, (2) broadcast at transplanting, (3) broadcast and incorporated at final harrowing, and (4) deep placement as sulfur-coated granules. The treatments were laid out in a randomized complete block design with four replicates. Measurements were done in the 1991 wet season, 1992 dry seaon (four treatments), and the 1992 wet season (only treatment 3). Methane emission rates from the experimental plots showed pronounced seasonal and diel variations. The diel pattern of methane emission rates followed a consistent pattern, with highest rates observed in the early afternoon and lowest rates in the early morning. Methane emission rate was generally highest at the ripening stage. The average methane emission rate during the 1992 dry season (190 mg CH4 m-2 d-1) exceeded the average flux rates of the 1992 wet season (79 mg CH4 m-2 d-1) by a factor of 2.4. The total methane emitted from these flooded rice fields amounted to 19 g CH4 m-2 in the dry season with rice yields of 5.2--6.3 t ha-1 and 7 g CH4 m-2 in the wet season with rice yields of 2.4--3.3 t ha-1 regardless of the mode of N application. Significant amounts corresponding to 20% of the methane released under waterlogged conditions were released when the soil was drained after harvest. Emission rates increased sharply when the floodwater receded and macropores started to drain. Emission of methane stopped only when the soil became fully aerated. ¿ American Geophysical Union 1994