Abstract: Sixty leaf samples of mature rice were scanned by ASD field pro3 for optical data and analyzed for As content using atomic absorption spectrometry. A space model for diagnosing As contamination in rice based on 12 hyperspectral potentially sensitive indices of As stress was used in this paper. In the first step, single diagnosis indices (PSNDa, DSWI, SIPI) representing different rice physiological parameters were determined using correlation analysis between hyper-spectral indices and As content, and then two-dimension diagnosis spaces (PSNDa-DSWI, PSNDa-SIPI, DSWI-SIPI) were constructed, of which PSNDa-SIPI representing chlorophyll and cell structure shows a more effective prediction of As contamination in rice. In the second step, another diagnosis space (F1-F2) was built from principal component analysis of the 12 hyperspectral indices, whose accumulative contribution rate was 88%. This was an exact predictor of As pollution in rice; where F1<1.95 and F2>0.75 represents high pollution, 1.95<F1<3.15 and 0.75>F2>0.40 represent medium pollution, and F1>3.15 and F2<0.40 represent low pollution. These diagnosis spaces form a comprehensive diagnosis space model that services for monitoring large-scale As contamination in rice from different levels.