Objective: 270 milk powder samples from 6 different brands were detected and distinguished by low field nuclear magnetic resonance combined with chemometrics. Methods: Three chemometrics methods of principal component analysis (PCA), partial least squares discriminant analysis (PLS-DA) and backpropagation artificial neural network (BP-ANN) were used to process experimental data of samples statistically. Results: The PCA method based on three-dimensional projection could not achieve the purpose of rapid identification of milk powder brand; the correct recognition rates of training and prediction sets were 66.1% and 52.2% for the PLS-DA method, respectively, which was low in credibility and challenging to realize the rapid identification of milk powder brand; the correct recognition rates of training and prediction sets of were 99.4% and 100.0% for the BP-ANN method respectively. Conclusion: The combination of low field nuclear magnetic resonance and BP-ANN can identify the milk powder brand well.