The nuclear magnetic resonance method (NMR) is widely used as a tool in the pore size characterization of rocks such as sandstones, carbonates, and coals. This technique also has been widely used to estimate the unfrozen water content in frozen soils; however, there are limited studies that focus on the frozen soils pore structure characterization. In the NMR tests, based on the results from the literature, a linear relationship is typically assumed between the pore size (R) and transverse relaxation time (T-2), of the transverse magnetization decay. This relationship is mathematically represented as 1/T-2 = 2 rho(2)/R, where rho(2) is a constant representing the surface relaxivity. Traditionally, rho(2) is determined by T-2 cutoff values, which is a relaxation time threshold that divides the T-2 spectrum into two zones: namely, the bound water and free water in frozen soils. Conventionally, centrifuge experiments are used for determining the cutoff value in the field of petroleum engineering. The key parameter rho(2) is also approximately estimated or assumed based on the information from the published literature for different types of soils. In other words, T-2 cutoff value is based on approximations that have limitations or based on centrifuge tests that need elaborate testing which is expensive. In this paper, a new method is proposed to calculate T-2 cutoff value from the NMR test results on a saturated frozen silt clay. The key advantage of this method is that it alleviates the approximations or the use of expensive centrifuge tests for the estimation of the cutoff value, T-2. The proposed method in this paper is useful for better understanding the behavior of frozen soils with the aid of simple and inexpensive methods.