Near infrared (NIR) spectroscopy technology based on a portable NIR analyzer, combined with kernel Isomap algorithm and generalized regression neural network (GRNN) has been applied to establishing quantitative models for prediction of acidity and vitamin C in six kinds of apple samples. The obtained results demonstrated that the fitting and the predictive accuracy of the models with kernel Isomap algorithm were satisfactory. The correlation between actual and predicted values of calibration samples (Rc) obtained by the acidity model was 0.999 4, and for prediction samples (Rp) was 0.979 9. The root mean square error of prediction set (RMSEP) was 0.055 8. For the vitamin C model, Rc was 0.989 1, Rp was 0.927 2, and RMSEP was 4.043 1. Results proved that the portable NIR analyzer can be a feasible tool for the determination of acidity and vitamin C in apples.