Abstract: For the detection and quantification of urea in milk, pure milk samples and 40 adulterated milk samples added different contents of urea were prepared. Then 2D correlation (NIR-NIR, IR-IR, NIR-IR) spectroscopy under the perturbation of adulteration concentration was calculated and the spectra in the range of 4 200-4 800 cm-1 and 1 400-1 704 cm-1 were selected to construct the partial least square (PLS) calibration model, respectively. The PLS calibration model showed 4 200-4 800 cm-1 was the better range for calibration performance and the root mean square errors of cross validation (RMSECV) of the model was 0.266 g/L. When using this model for predicting the urea contents in prediction set, the root mean square errors of prediction (RMSEP) was 0.219 g/L and the coefficient correlation of actual values and predicted values was 0.999, which means the model has good prediction ability. The method can be used for a correct discrimination on whether the milk is adulterated and provides a new and cost-effective alternative to test the adulteration of milk.