A fast determination method was established for content determination of mashed potato in noodles. Studied 170 potato noodles by near infrared spectroscopy and stoichiometry to establish prediction model by PLS. The result indicated that the prediction model based on wave number of 9 403.6～5 446.2 cm^-1, and spectral pretreatment by the standard normal variate transformation (SNV) showed an excellent prediction accuracy. The external validation determinants was 0.956 4, the root mean square error of prediction was 3.73%, the slope was 1.029, and relative prediction deviation was 2.04. The results indicate that near infrared spectroscopy can be used as an efficient way to detect the mashed potatoes content of potato noodles.