A method for evaluating the overall uncertainty of alanine EPR transfer dosimetry in the therapeutic dose range is described. The method uses experimental data on EPR signal reproducibility from replicate dosimeters irradiated to low doses (1-5 Gy), estimates of Type B uncertainties, and Monte Carlo simulations of heteroscedastic orthogonal linear regression. A Bruker ECS106 spectrometer and Bruker alanine dosimeters have been used for this evaluation. The results demonstrate that alanine dosimetry can be used for transfer dosimetry in that range with the overall uncertainty 1.5-4% (1sigma) depending on the dose, the number of replicate dosimeters. and the duration of the calibration session (the session should not exceed one working day).