Abstract:A portable visible (Vis)/near infrared (NIR) spectrometer was performed on pears (Pyrus communis L.) cv ‘Abate’, ‘Conference’ and ‘Wujiuxiang’ for firmness assessment by diffused transmittance technique. One set of 96 pears were used to explore the performance of this technique on developing global firmness models with partial least squares (PLS) and multiple linear regression (MLR) methods. The spectra were treated by 2nd order derivation with Savitsky–Golay (polynomial order 2, data point 25). Genetic algorithms partial least squares (GA-PLS) and regression coefficient method were used to select informative variable for PLS calibration. The global models were obtained by PLS to predict firmness yield the root mean standard error of prediction (RMSEP) of 7.780N and 8.080N, and the relative standard deviation of prediction (RSD) of 26.24% and 29.71%, respectively. The global model for firmness measurement was also obtained by MLR with RMSEP of 7.740N and RSD of 26.10%, respectively. The results highlight the potential of portable Vis/NIR instruments for assessing internal quality of European pear on-site.