The efficient prediction of freezing processes is important for the frozen food industry with the purpose of preserving food quality. A comparison is made between the finite difference and finite volume methods to predict the unsteady 2D temperature distribution for the freezing of three different ground meat cylinders, with variable convective boundary conditions. Experiments were carried out to measure the local heat transfer coefficients as a function of freezing time and location. Time-dependent air temperature conditions caused by thermostat action were included in the analysis. The meat thermophysical properties as density, specific heat, and thermal conductivity are assumed to vary non-linearly with temperature. The experimentally measured time-temperature data were compared with the simulation results obtained by both numerical methods. According to statistical test results (RE %), the finite volume method gave the best fit quality. The finite volume method is more accurate for simulating the freezing process, as well as estimating the freezing time of this product