The Haugh unit, yolk index and weight loss rate are important indicators to evaluate the freshness of eggs. Near-infrared (550~985 nm) transmission spectra expressing the egg internal quality are collected to make non-destructive quantitative analysis. Pretreatment of spectrum is based on Savitzky-Golay spectrum smoothness method and first-order differential. A model is established using synergic-interval partial least square (SI-PLS) method for an appropriate spectral or wavelength range. Effective information of the spectrum, obtained from the principal component analysis of characteristic variables, is input into the back propagation artificial neural network (BP-ANN) to set a nonlinear model. The comparison of the two models shows that the SI-PLS model is better than the BP-ANN model. The results prove that the application of near-infrared transmission spectrum to evaluate egg freshness is feasible.