A nondestructive, efficient, and rapid method for quantitative analysis of two bioactive components (salidroside and p-tyrosol) in Rhodiola crenulata, a traditional Tibetan medicine, by Fourier transform near-infrared (FT-NIR) spectroscopy was developed. Near-infrared diffuse reflectance spectra in the range of 4000 to 10000 cm⁻¹ of 50 samples of Rhodiola crenulata with different sources were measured. To get a satisfying result, partial least squares regression (PLSR) was used to establish NIR models for salidroside and p-tyrosol content determination. Different preprocessing methods, including smoothing, taking a second derivative, standard normal variate (SNV) transformation, and multiplicative scatter correction (MSC), were investigated to improve the model accuracy of PLSR. The performance of the two final models (salidroside model and p-tyrosol model) was evaluated by factors such as the values of correlation coefficient (R²), root mean square error of prediction (RMSEP), and root mean square error of calibration (RMSEC). The optimal results of the PLSR model of salidroside showed that R², RMSEP and RMSEC were 0.99572, 0.0294 and 0.0309, respectively. Meanwhile, in the optimization model of p-tyrosol, the R², RMSEP and RMSEC were 0.99714, 0.0154 and 0.0168, respectively. These results demonstrate that FT-NIR spectroscopy not only provides a precise, rapid method for quantitative analysis of major effective constituents in Rhodiola crenulata, but can also be applied to the quality control of Rhodiola crenulata.