AbstractCarbon dioxide (CO₂) and methane (CH₄) are the most important two greenhouse gases in the atmosphere which were closely coupled between terrestrial ecosystems and the atmosphere. However, the relationship between CO₂ and CH₄ fluxes of rice paddy at different temporal scales is still unclear. Based on 6 years of eddy covariance measurements on a flooded rice paddy field in Eastern China, the relationship between gross primary productivity (GPP) and CH₄ fluxes and the effects of biophysical and environmental factors on daily CH₄/GPP were investigated. CH₄ fluxes and GPP were tightly linked at diurnal and seasonal scales. On average, half‐hourly CH₄ fluxes lagged GPP by 1.5 hr while daily GPP lagged CH₄ fluxes by 19 days over the growing season. Leaf area index, water table depth (WTD), air temperature, vapor pressure deficit, and canopy conductance significantly affect daily CH₄/GPP. A semi‐empirical multiplicative model was able to capture 77% and 84% seasonal variability of daily CH₄/GPP for the training data set and testing data set, respectively. WTD is a crucial input affecting the performance of the model. Without WTD included, the agreement between estimated and observed daily CH₄/GPP was weakened. Under the condition without WTD data available, separate calibration of the semi‐empirical multiplicative model before and after panicle initiation could improve the estimation of daily CH₄/GPP to some extent. The findings in this study enhance our understanding regarding the complex interactions between CH₄ fluxes and GPP in the rice paddy and are helpful for better estimation of CH₄ fluxes from GPP.