In order to minimize the great influence of the distance between fiber-optics probe and sample surface point on prediction results in livestock products quality and safety detection, the spectral change rules with change distances, and proposed a calibration method for the collected spectra were analyzed. Pork tenderness was investigated, and the calibration result of the proposed method was verified. First, VIS/NIR spectral data of pork (longissimusdorsi muscle) was collected by a fixed probe with the changing distance between 4 and 18mm. Then, the spectral data collected under each condition were preprocessed by the same method and PLSR prediction models were established. The correlation coefficient and root meansquare error of validation set were 0.82 and 5.82N respectively with unadjusted probe. Between 380nm and 1050nm, all the fitting coefficients obtained were higher than 0.999 with root mean square errors in the range of 0.0187~0.2422 when the spectral data at the distance of 4~18mm were calibrated to 13mm by fifth-order polynomial fitting. The correlation coefficient and root mean standard error for pork tenderness prediction were in the range of 0.83~0.90 and 4.80~5.75N respectively when the distance was at 4~17mm, by using the fitted spectral data and the best prediction model.The results show that the proposed method can overcome the influence of different distances on pork quality prediction by calibrating the spectral data.