Visible/near infrared (Vis/NIR) spectroscopy was used to determine soluble solid content (SSC) of navel oranges with pericarp and without pericarp, and the effect of pericarp on prediction accuracy of SSC of navel oranges was investigated. In addition, Vis/NIR spectra of navel oranges with pericarp and without pericarp were acquired by a QualitySpec spectrometer in the wavelength range of 350~1 000 nm, and the effect of pericarp was analyzed from two aspects of spectrum and model performance. The average spectra of navel oranges with pericarp and without pericarp were compared, and 20 principal components that obtained were used for multivariate analysis of variance (MANOVA). Moreover, partial least squares (PLS) regression combined with different pretreatment methods was used to develop calibration models of SSC for navel oranges with pericarp and without pericarp. Furthermore, the performance of models was compared, and square of prediction residuals of samples in prediction set were used for analysis of variance (ANOVA). The results indicate that the effect of pericarp on prediction accuracy of soluble solid content in navel oranges is significant at 5% confidence level. The correlation coefficients of prediction set and root mean square errors of prediction (RMSEPs) of PLS of SSC for navel oranges with pericarp and without pericarp are 0888, 0456% and 0944, 0324%, respectively.