A single indicator is traditionally used for assessing the quality and efficacy of traditional Chinese medicine (TCM), which is insufficient due to the complexity of TCM. Multiple indicators detection are necessary for quality control of TCM. The feasibility for simultaneous quantification of baicalin, alcohol-soluble extractives, moisture, and ash in Radix Scutellariae is investigated by near-infrared (NIR) spectroscopy combined with chemometrics. First, random sampling, concentration sorting, Kennard-Stone (KS), Duplex, and sample set partitioning based on joint X-Y distances were investigated for grouping of training and validation sets. Then, five multivariate calibration methods were compared for model establishment between NIR spectra and the four components including principal component regression, support vector regression, partial least squares regression (PLSR), back propagation neural network, and extreme learning machine. Meanwhile, different preprocessing methods, such as SG smoothing, 1st derivative, 2nd derivative, continuous wavelet transform, standard normal variate, multiplicative scatter correction, and their combinations were further investigated to improve the performance of the optimal models. Uninformative variable elimination, Monte Carlo uninformative variable elimination, randomization test, competitive adaptive reweighted sampling, genetic algorithm, whale optimization algorithm, and butterfly optimization algorithm (BOA) were finally adopted to select relevant variables. Results show that the KS, PLSR and BOA are the best data grouping, calibration and variable selection methods for four components, respectively. The determination coefficient of the optimal model for baicalin, alcohol-soluble extractives, moisture, and ash in Radix Scutellariae are 0.975, 0.997, 0.987 and 0.891, respectively, indicating the efficiency of chemometrics with NIR for simultaneous quantification of multiple components in TCM.