Abstract: To study the changes of chlorophyll relative content（SPAD） in rice leaves under three water treatments and five nitrogen treatments, and to explore the feasibility of multispectral remote sensing technology for unmanned aerial vehicle（UAV） to retrieve rice SPAD. In this study, DJI Phantom 4 multispectral UAV was used to collect multispectral remote sensing images of rice canopy at jointing and booting stage, heading and flowering stage and milk ripening stage, and simultaneously measure the SPAD value of rice. Based on twenty-five spectral variables（five band reflectance and twenty vegetation indexes）, the retrieving model of rice SPAD at different stages is established by multiple linear stepwise regression（MLSR）, ridge regression（RR） and lasso regression（LR）. The results show that the best retrieving models of rice SPAD at three growth stages are established by lasso regression, and the best retrieving model of SPAD established at milk ripening stage has the highest inversion accuracy among the three growth stages, with coefficient of determination of 0.782, root mean square error of 1.217 7 and relative error of 6.611 3%. Therefore, this paper can monitor rice leaf SPAD by remote sensing, and provide a scientific basis and data support for rice precision irrigation and fertilization.