Genetic algorithms (GAs) are used to implement an automated wavelength selection procedure for use in building multivariate calibration models based on partial least squares regression. The GAs also allows the number of latent variables used in constructing the calibration models to be optimized along with the selection of the wavelengths. This method was applied to fundamental study of non-destructive measurement of intact fruit quality with Fourier transform near infrared spectroscopy (FT-NIR). The experiments tested in this method are sugar content, titratable acidity and valid acidity. The optimal configurations for the GAs were investigated for each data set through experimental design techniques. Despite the complexity of the spectral data, the GA procedure was found to perform well (RMSEP=0.395, 0.0195, 0.0087 for SC, TA and pH respectively), leading to calibration models that significantly outperform those based on full spectrum analyses (RMSEP=0.512, 0.0198, 0.0111for SC, TA and pH respectively). In addition, a significant reduction in the number of spectral points required to build the models is realized and all of the numbers of wavelengths for building the models can reduce by 84.4%. It is instructive for the further study of the theory of non-destructive measurement of the fruit internal quality with FT-NIR spectroscopy.