Cerebral edema is a secondary symptom of several conditions, including stroke, liver, failure, and severe craniocerebral injury. This study aimed to propose a Low-Field Nuclear Magnetic Resonance (LF-NMR) method for rapid and nondestructive measurement of brain water content and water phase state evaluating animal brain edema models. The water transverse relaxation time (T2) distribution in brain tissues was ascertained by measuring the T2 spectrograms of mouse brain tissues following various drying durations, in accordance with the shifting law of T2 peaks. A methodological study was carried out for the determination of water content in mouse brain tissue by the LF-NMR Peak Area (LF-NMR-PA) method using CuSO4 solution as standard solution. This method was studied in comparison with the wet/dry weight method and the Partial Least Squares Regression (PLSR) prediction method in normal and MCAO model mice, and further applied to the study of mannitol treatment of cerebral edema to validate the precision and accuracy of the hydration measurements as well as the method’s impact on subsequent experiments of cerebral infarct area measurement. In the T2 spectra of mouse brain tissues, the separate peaks T21 (0–10 ms), T22 (10–100 ms), and T23 (100–1000 ms) correspond to fat and bound water, water, and free water in brain tissue, respectively. The LF-NMR-PA method had good specificity, linearity, precision, stability, reproducibility, and recovery. The accuracy and range of use of the LF-NMR-PA method were better than those of the PLSR method, and its results were not significantly different from those of the wet/dry weight method. However, the short measurement time of the LF-NMR-PA method ensured the integrity of the tissues, and the determination of the moisture by this method did not have a significant effect on the subsequent determination of the area of cerebral infarction, which ensured the consistency of the experimental results.