Aronia melanocarpa (Michx.) Elliott. It is a berry from the Rosaceae Family rich in polyphenols, known as its main chemical components, including anthocyanins, flavonoid glycosides, tannins, etc., of A. melanocarpa. It has shown antioxidant, bacteriostatic, anti-tumor, anti-inflammatory, weight loss, glucose regulation, lipids, and other pharmacological activities. It has now been added to the list of new raw food materials. The polyphenol content of A. melanocarpa is closely related to its efficacy value. Therefore, improving their detection method is crucial to standardizing the raw material and product market from A. melanocarpa. However, the current detection method is cumbersome and time-consuming, and it is difficult to meet the industrial development needs of A. melanocarpa after it enters the list of new food raw materials. Thus, It is urgent to develop a method for rapidly determining polyphenol content. Mid-infrared spectroscopy established a rapid and quantitative determination method of polyphenol content in A. melanocarpa. The infrared spectral data of 750 samples from A. melanocarpa in 15 regions were collected for the spectral analysis, and the content of polyphenols in each sample was measured. The K-S sample division method was used to divide the sample into a correction set and verification set in the proportion of 4∶1. The grouped spectral information was pretreated by multiple scattering correction (MSC), standard normalization (SNV), smoothing (SG), first derivative (FD), second derivative (SD) and other spectral preprocessing methods. Compared with the original spectrum by random forest regression (RFR) modeling and prediction, the best spectral preprocessing method was determined as MSC. The competitive adaptive reweighting algorithm (CARS) and continuous projection algorithm (SPA) were used to select the optimal characteristic spectral wavelength of the polyphenols of A. melanocarpa. The spectral data selected by the two methods were combined with random forest regression (RFR), partial least squares regression (PLSR), limit learning machine (ELM), and support vector machine regression (SVR) for modeling and comparison to determine the optimal algorithm model. The results showed that the CARS algorithm can effectively reduce the redundancy of infrared spectral data and improve the accuracy and stability of model prediction. The CARS-RFR model had the best prediction performance. Its correction set R_c, RMSEC, verification set R_p, RMSEP, and RPD were 0.986 5, 0.073 2, 0.974 3, 0.100 6, and 6.235 6, respectively. The above results revealed that the combination of mid-infrared spectroscopy and chemometrics, especially the CARS-RFR model, can effectively, rapidly, and accurately detect the polyphenol content of A. melanocarpa. The research results can thus provide technical support for rapidly determining the polyphenol content of A. melanocarpa.