Rice paddies are a major source of atmospheric methane (CH4). However, the strength of this source is highly uncertain. To improve estimates of CH4 emission at the regional and global scales requires process-based models, integrating the environmental and biological factors that determine the rate of CH4 emission. In this study, we used such a model to calculate the spatial and seasonal variations in CH, emission from rice paddies, with the support of an integrated GIS data set defining distribution of rice paddies, rice calendar, climate, and soil conditions. The calculated rates of CH4 emission in a rice-growing season varied greatly, from 5 to 90 g CH4 m(-2), for rice paddies across the globe. In most rice paddies daily mean CH4 effluxes were from 50 to 400 mg CH4 m(-2). Annual total emission was estimated to be 53 Tg; two thirds of this was emitted between 10 degrees N and 30 degrees N. Monthly CH4 emissions ranged from 1.4 Tg in January to 8.7 Tg in August, and more than a half of the emission occurred between June and October. Our model suggests that, on average, 12% of the simple carbon substrate from decomposition of soil organic matter and rice plants was transformed into CH4, and the emitted CH4 accounted for 6.5% of the rice net primary production. Copyright (C) 1996 Elsevier Science Ltd