Rice fields have always been regarded as one of the largest anthropogenic sources of atmospheric methane. Here we report the results of a 7-year study of methane emissions from rice fields in the Sichuan Province of China. In this region, there is one crop of rice per year, the fields are continuously flooded from transplanting to harvest, and there is heavy use of organic fertilizers. Emissions over the entire growing season were measured from each of up to 24 plots. Environmental variables were measured and relevant supporting data on the agricultural practices were recorded. The fields were studied under prevailing agricultural practices of the local farmers. The results represent emissions under standard agricultural practices and the year to year variability of climate, fertilizers, available irrigation water, and cultivars. Based on some 5000 flux measurements, the average emission rates between 1988 and 1994 were 30 mg/m2/h for a growing season of between 100 and 120 days. This emission rate is comparable to other published data from similar rice fields but somewhat on the high side of the range. There were no systematic trends of emissions during the 7 years of our experiment, but there was substantial year to year variability. The data have been subjected to exhaustive analyses for validity, accuracy, and reliability. From this, a high-quality, spatially averaged data set has been constructed representing average emissions from the rice fields for each day when measurements were taken. We describe here the main observational results and document the spatial and temporal variability observed on timescales ranging from a day to several years and on spatial scales ranging from 0.5 m2 to 16 m2. ¿ 1998 American Geophysical Union