Atherosclerotic carotid stenosis (ACS) is a recognized risk factor for ischemic stroke. Currently, the gold diagnostic standard is Doppler ultrasound, the results of which do not provide certainty whether a given person should be qualified for surgery or not, because in some patients, carotid artery stenosis, for example at the level of 70 %, does not cause ischemic stroke in others yes. Therefore, there is a need for new methods that will clearly indicate the marker qualifying the patient for surgery. In this article we used Fourier Transform InfraRed Attenuated Total Reflectance (FTIR-ATR) spectra of serum collected from healthy and patients suffering from ACS, which had surgery were analyzed by machine learning and Principal Component Analysis (PCA) to determine chemical differences and spectroscopy marker of ACS. PCA demonstrated clearly differentiation between serum collected from healthy and non-healthy patients. Obtained results showed that in serum collected from ACS patients, higher absorbances of PO2− stretching symmetric, CH2 and CH3 symmetric and asymmetric and amide I vibrations were noticed than in control group. Moreover, lack of peak at 1106 cm−1 was observed in spectrum of serum from non-control group. As a result of spectral shifts analysis was found that the most important role in distinguishing between healthy and unhealthy patients is played by FTIR ranges caused by vibrations of PO2− phospholipids, amides III, II and CO lipid vibrations. Continuing, peaks at 1636 cm−1 and 2963 cm−1 were proposed as a potential spectroscopy markers of ACS. Finally, accuracy of obtained results higher than 90 % suggested, that FTIR-ATR can be used as an additional diagnostic tool in ACS qualifying for surgery.