The potential of low field NMR (LF NMR) as a fast monitoring technique to estimate the quality of hake (Merluccius merluccius) frozen stored at −10 °C for up to 6 months was evaluated. LF NMR clearly detected three populations of water: water strongly bound to macromolecules (T2b), trapped water (T21) and free water (T22). As storage time increased, and concomitant with an increase in the T22 and a decrease in the T21 water populations, the water holding capacity (WHC) and apparent viscosity values decreased and the shear strength increased, reflecting the characteristic loss of juiciness and tougher texture developed by hake during frozen storage. Two mathematical models were constructed: a simple regression using the biexponential analysis of the relaxation times (T21, T22) and amplitudes (A21, A22) and a partial least square regression (PLS) of CONTIN analysis. Both models seemed suitable to estimate the quality of the product.