Rice quality is directly related to human health, so it is important to have traceability systems that can trace inferior or contaminated rice back to its geographical origin. This ensures farming practices in substandard regions become better regulated to improve rice quality, origin labelling and consumer trust. However, tracing the origin of rice on the marketplace requires an accurate database benchmarking the isotope distribution over areas of rice production. Large stable isotope data sets can be used to determine the geographical origin of rice through predictive isoscape models. This study presents the first rice isoscape based on environmental similarity to predict the geospatial distribution of δ13C, δ2H and δ18O values of Chinese rice and provides uncertainty at every location such prediction is made. For this study, 794 rice samples were collected in 2017 from primary rice production regions of China. An independent verification shows that the predicted isotope distribution from this new approach is of high accuracy, with a root mean squared error (RMSE) of 0.51‰, 7.09‰ and 2.06‰ for δ13C, δ2H and δ18O values respectively. In addition, uncertainty in the spatial distribution of isotopes can be used to indicate the prediction accuracy and to guide future sampling. Our results indicate that an isoscape prediction method based on environmental similarity is effective to predict the spatial distribution of stable isotope in rice, and is an effective tool for building isotope distribution in rice over large areas with complex environment. This method could also be used to predict potential isotopic variations in future years due to climate change.