A new data analysis procedure is proposed for obtaining unsteady wall shear stress from flush-mounted hot-film anemometer measurements. The method is based on a two-dimensional heat transfer model of the unsteady heat transfer in both the hot-film sensor and in the gas flow. The sensor thermal properties are found from preliminary calibration experiments at zero flow. Numerical experiments are used to demonstrate the data analysis method using simulated sensor signals that are corrupted with noise. The numerical experiments show that noise in the data propagates into the results so that data smoothing may be important in analyzing experimental data. Because the data analysis procedure is linear, a linear digital filter is constructed that could be used for processing large amounts of experimental data. However, further refinements will be needed before the method can be applied to experimental data.