Herba Epimedii contains high amounts of flavonoids, such as epimedin and icariin, which are efficient in tonifying kidney and improving immunity. Nowadays, various chemical analysis methods have been applied to measure the flavonoid content of Herba Epimedii. However, these traditional methods are destructive, time-consuming, and costly and cannot meet the requirements of massive samples analysis in pharmaceutical production and plant breeding. As a rapid and effective tool for quantitative determination and process monitoring, modern hyperspectral analysis technology has earned more and more concerns. However, for the full-range spectra, the existence of insignificant and irrelevant spectral variables can weaken the calibration models' accuracy and efficiency. Therefore, the spectral variables selection is essential to improve the performance of the final models by eliminating the uninformative bands. In this study, the partial least squares regression (PLSR) coupled with the genetic algorithm (GA) variables selection procedure, namely GA-PLSR, was used to estimate epimedin A, epimedin B, epimedin C, and icariin content in Herba Epimedii. This paper aims to explore the feasibility of hyperspectral analysis technology in the measurement of the pharmacologically active constituents in Herba Epimedii and further explore their important spectral response bands. The results show thatthe hyperspectral analysis technology combined with chemometrics exhibited considerable potential for rapid and nondestructive assessment of Herba Epimedii. When compared with full-spectrum PLSR models, GA-PLSR models could improve the accuracies and robustness of epimedin A, epimedin B, epimedin C, and icariin content measurements (with RCV2 values increased from 0.645, 0.720, 0.718, and 0.642 to 0.671, 0.835, 0.782, and 0.796, and with RMSE_CV values declined from 2.102, 2.896, 21.069, and 1.221 to 2.071, 2.230, 18.656, and 0.912, respectively). Besides, we found some feature wavelengths, mainly around 690~740 and 420 nm, which play important roles in detecting pharmacologically active constituents in Herba Epimedii. Given these desirable findings, this study can provide a valuable reference for the rapid and accurate measurement of epimedin A, epimedin B, epimedin C, and icariin contents by hyperspectral technology, can provide a theoretical basis for the design of spectral sensors in qualifying Herba Epimedii.