Verification of advanced radiotherapy treatment modalities requires measurement of three-dimensional absorbed dose distributions with high spatial resolution and precision. Polymer gel dosimeters combined with magnetic resonance imaging may be able to fulfil this requirement. However, verification requires that the uncertainty in the dosimeter is well known. One method of estimating the overall uncertainty in polymer gel dosimeters involves the propagation of the uncertainty in the R2 (nuclear magnetic resonance relaxation rate) map and the uncertainties in the calibration data. This work shows that using this method with current data suggests that the lowest uncertainty currently obtainable is about 3% at 8 Gy and 7% at 2 Gy. Furthermore, the most significant reductions in overall uncertainty will be achieved by reducing the noise in the R2 map.