In order to accurately monitor and objectively evaluate the damage of the rice leaf roller, Cnaphalocrocis medinalis Guenee (C. medinalis) , to rice growth, breeding and yield formation, in this paper, the canopy hyperspectral data and SPAD (Soil and plant analyzer development) values of rice in the controlled field experiments (in 2015 and 2019) and the natural field experiments (in 2020) were measured by the instruments of ASD Field Spec3 and SPAD-502 at different growth stages (jointing stage, booting stage, filling stage and maturing stage), and the pest number and the percentage of rolled leaves in rice resulted from C. medinalis was measured by the manual investigations. The hyperspectral characteristics of rice canopy, physiological and ecological parameters of rice and the characteristics of C. medinalis’ occurrence parameters in the two experiments were analyzed. The models for estimating physiological and ecological parameters of rice after being damaged by C. medinalis based on the hyperspectral parameters was established. The results were showed as follows: (1) In the two experiments, the reflectance of the red edge to near-infrared band of rice canopy and SPAD value decreased with the aggravation of the degree of C. medinalis infestation but the reflectivity of the visible light band was the opposite. (2) The SPAD value of the natural field experiments and their canopy reflectance from the red light to near-infrared wavelength range were significantly lower than those of the control field experiments in the early two stage of rice growth and development, but slightly higher than that of the control field test in the latest two stages. (3) After comprehensively analyzing the experimental data, the multiple pest characteristic parameters and vegetation indices in the natural field experiments and the controlled field experiments were selected, and a set of single factor estimation models and multi-factor estimation models of SPAD were constructed respectively. Each model achieved the good estimation effect, among which the binomial function simulation effect of EVI was the best in the single factor models, however, the simulation effect of the multi-factor linear regression estimation model was better than that of all single-factor models. (4) Through the application test of these models in 2021, it was found that the estimated SPAD values of the single factor estimation model based on pest population, leaf curling rate, OSAVI, EVI and DVI and the multi factor estimation model based on RVI and leaf curling rate were highly consistent with the measured values among these models and achieved a relatively ideal estimation effect . This study has provided a high-precision and feasible estimation method for the SPAD value estimation of rice under the infestation of C. medinalis.