Milk contaminated by chemicals is considered adulterated and inadequate for consumption, representing a risk to consumer health. Among these contaminants, great importance is given to the detection of antibiotic contamination to be used in the treatment of other infections and mastitis in dairy cows. Antibiotics may also be added to milk voluntarily to increase its durability. In this work, we prepared 10 samples of sterilized milk (UHT) with concentrations from 0.002 to 1% of the antibiotic Amoxicillin trihydrate. The aim was to evaluate the concentration low limit of antibiotic that may be detected by the method of Multivariate Curve Resolution (MCR) applied to mid-infrared spectra in Fourier transform (FT-MIR). The spectra were obtained in triplicate using a SpectrumTwo spectrometer (Perkin Elmer) with the following settings: scan 4000-600 cm-1, resolution 4 cm-1 and 16 scans. The spectra obtained had their baselines corrected using the standardized baseline. The data set was applied MCR using ALS toolbox 7.1 and MATLAB 7.0 (MathWorks Inc.). The relationship between the concentrations matrix generated by the MCR and the array of tampering by the antibiotic concentrations was linear at concentrations of 1% to 0.13804%. In lower concentrations it was observed an overlap of spectra that made the identification of presence of Amoxicillin not able to be detected. It was concluded the methodology can be used to detect the presence of antibiotics in milk since the concentrations are greater than 0.138%.