The objective of this research was to investigate the relationship between the degree of milling and surface lipid content of white rice. The sample was the 98-2 TK No.9 rice from Taichung District Agricultural Research and Extension Station. The rice sample with approximately 13% moisture was dehulled with a Long-Good dehuller（THM-1）. The brown rice was milled with a laboratory no. 2 McGill mill to provide 8 different degree of milling of samples. The sample was put in Ziploc Storage Bags and stored at 25℃. The whiteness was measured by Color Tester LFM1, and surface lipid content was extracted through Soxhlet extraction. The near-infrared absorbance spectrum for whole kernel rice was obtained from FOSS NIRS 6500. The coefficients of determination （R2） of regression of degree of milling were 0.98 and 0.99 for whiteness and surface lipid content respectively. The difference of absorbance for different samples was more clear after 1400nm. The wavelength selected by step-wise regression were 1436nm and 806nm, and the determination （R2） of regression for prediction equation is 0.99. In the simulation study, LED light source L1450-66-60 was chosen from epitex, and the sensor was PD1450-35T52. The coefficients of determination （R2） of regression of prediction current were 0.97、0.96 and 0.96 for degree of milling、whiteness and surface lipid content respectively that suggested the feasibility to develop a whiteness measurement instrument using NIR LED as the light source.