Nondestructive determination of sugar content (SC) of fragrant pears with temperature variation （5℃，10℃，15℃ and 20℃）was studied by visibleNIR spectroscopy coupled with multivariate calibration methods. In the region of 500~900 nm, five multivariate calibration models: stepwise multiple linear regression (SMLR), partial least squares (PLS), least squaressupport vector machines (LS-SVM1, only 30 bands spectral data obtained by SMLR analysis inputted as X variable), LS-SVM2 (both 30 bands spectral data and sample temperature inputted as X variable) and genetic algorithmpartial least squares (GA-PLS) were compared. The results showed that the prediction performance was in the order of LS-SVM2, LS-SVM1, GA-PLS, PLS and SMLR.