The objective of this paper was to study the potential of visible/near infrared reflectance spectroscopy (Vis/NIR) for nondestructive determination of potassium (K) content in fresh pear leaves with an ASD FieldSpec 3 spectrometer. All the samples were divided randomly into two groups, one with 120 samples as the calibration set, and the other with 30 samples as the validation set. Different spectra wave bands, spectra preprocessing in different ways and different spectra styles were used in the prediction model of K content. The results showed that, after the preprocessing of original spectral with the whole wave band plus S-G (3) smoothing, the prediction model calibrated with 17 principal component factors had the best performance by the partial least squares (PLS) regression, in which the determination coefficient (R2) of calibration and cross-validation was 0.722 7 and 0.679 1, respectively, and the root mean square error of cross-validation (RMSECV) was 1.171. Test of the best PLS model with 30 samples in the validation set showed that the predicted average relative error was 6.81%. Therefore, it could be concluded that Vis/NIR has a huge potential for the determination of total potassium content in fresh pear leaves in a rapid and nondestructive way.