There are many methods to test the water content, pore water distribution and pore structure characteristics of cement-based materials, and they are different in the measurement principle, accuracy, ease of use and reliability. As a non-destructive testing method, low-field nuclear magnetic resonance (LF-NMR) relaxation technology uses hydrogen and ¹H nuclei in water as a probe. By testing the relaxation signal of ¹H nuclei, it can accurately detect the content of hydrogen protons and the interaction with the pore wall, thereby indirectly obtaining features such as water content, water distribution in pores of different sizes and pore size distribution. They can be tracked and monitored over time and through other physical and chemical processes. Compared to traditional methods such as weighing method for water content and mercury intrusion porosimetry or nitrogen adsorption for porous structure measurement, LF-NMR relaxation technology can perform non-destructive testing without drying. It avoids the destruction of the abundant nanopore structure caused by drying pretreatment, and the pore measurement range covers the nanometer and micron scale pores. These significant technical advantages are very key to the study of the properties of cement-based materials. Because of the undisturbed, non-destructive and accurate characteristics of LF-NMR relaxation technology, it is precise, and thus more and more widely used in the field of cement-based materials, especially the hydration of cement, volume stability, actions of drying-wetting and freezing-thawing cycles, and durability performance. Since almost all these studies are closely related to water content and pore structure, LF-NMR plays an increasingly key role.Based on the relaxation mechanism of ¹H nuclei and the commonly used relaxation measurement methods, this paper systematically describes the transverse relaxation mechanism of pore fluid and the inversion algorithm of transverse relaxation data. The determination methods of surface relaxivity and quantitative calibration of transverse relaxation signal and continuous/discrete relaxation time spectrum into water content and continuous/discrete pore size distribution are described in detail. In addition, the application status of LF-NMR is summarized and prospected from 5 different perspectives: quantitative characterization of cement hydration process, pore structure at saturation, pore-scale water distribution at unsaturated state, imaging of water spatial distribution, and freezing-thawing damage process analysis.First of all, by testing relaxation magnetization and transverse relaxation time spectrum, LF-NMR resonance technology can non-destructively monitor the moisture content and its physical constraint state in the hydration process of cement at real time, so as to analyze the time-varying process of pores, especially nanopores, and apprehend the development history of microstructure, so as to realize the undisturbed characterization and analysis of cement hydration process. Secondly, the porosity and pore size distribution of hardened cement-based materials can be obtained according to the relaxation magnetization and relaxation time spectrum, allowing for qualitative and quantitative investigation of the relationship between pore structure and macroscopic properties. Thirdly, the pore classification method based on discrete or distinct peak T₂ spectrum can monitor the distribution of water in pores of different sizes at real time and continuously under unsaturated conditions, which can be used to study the macroscopic properties of cement-based materials such as moisture shrinkage. Fourthly, the water content and relaxation time spectra of different locations can be obtained by one-dimensional imaging, which can directly characterize the internal water migration process of cement-based materials. Finally, since the LF-NMR relaxation technique can only monitor the ¹H nuclear signal in evaporable water other than in solid phase such as ice and calcium hydroxide, it can be used to detect the phase transition of liquid water into ice.Summary and prospectsIt is found after literature reviewing that the LF-NMR relaxation technique using ¹H nuclei in pore fluid as a probe is especially suitable for characterizing cement hydration, pore structure evolution and pore-scale water allocation and spatial distribution of cement-based materials. On the basis of reasonable selection of hardware platform, test method, test parameters, inversion algorithm and accurate calibration of relaxation magnetization and transverse relaxation time, LF-NMR technology can help accurately measure important information closely related to pore structure and pore-scale water allocation of cement-based materials, such as hydration degree, pore size distribution and unsaturated pore water distribution non-destructively.Due to the influence of ferromagnetic substances, current LF-NMR analysis is mainly recommended to use white cement as the principal cementitious material. In future, it is necessary to fully consider the influence of ferromagnetic materials on transverse and longitudinal relaxation, to establish LF-NMR relaxation test method suitable for ordinary Portland cement, and to improve the signal-to-noise ratio of relaxation signals, to reduce the dead time of the probe, and to develop an inversion algorithm with good accuracy and precision. The probe diameter (sample chamber size) should be increased to support the test analysis of concrete specimen.LF-NMR relaxation technology is the promising characterization method that can detect the multi-scale pore structure and unsaturated pore water allocation and other key information of cement-based materials without any pretreatment. Since most of the key properties of cement-based materials, such as volume stability and durability, are closely related to pore structure and pore-scale water allocation, LF-NMR relaxation technology is expected to be able to drive breakthroughs and promote the development of modern concrete science and technology, similar to the encouraging role of magnetic resonance imaging in clinical diagnosis.