Genetic algorithm is widely used in NIRS data optimization, which is not limited by searching space. The region selecting by genetic algorithms (R-SGA)was applied in building calibration model of soluble solid content (SSC)of Pyrus pyrifolia, and the number of variables used to build calibration was further reduced from 2 075 to 690 in all of 3 models. Studies were performed to build GA-PLS models by different R-SGA latent variables, and the optimal R-SGA latent variables of Hosui, Wonhwang and Whangkeumbae pear were 10, 12 and 16, respectively. The R-SGA procedure was found to perform well (RMSEP = 0.608 and 0.524 for Hosui and Whangkeumbae pear respectively), leading to calibration models that significantly outperform those based on full-spectrum analyses (RMSEP = 0.632, 0.540). The prediction precision of GA-PLS models was similar to FULL-PLS for Wonhwang pear, with RMSEP of 0.610/0.595. In addition, the selected regions from R-SGA methods were used to build mixed model of 3 Pyrus pyrifolia varieties. The results indicated that the prediction precision of GA-PLS model was close to that of the full spectrum model, with RMSEP of 0.641 and 0.645, respectively. This work proved that the R-SGA could find optimal values for several disparate variables associated with the calibration model and that the PLS procedure could be integrated into the objective function driving the optimization, and it was feasible to build a universal model of different Pyrus pyrifolia varieties.