Ferrous‐sulfate‐doped gelatin gel dosimeters are useful tools for the measurement of three‐dimensional absorbed radiation dose distributions. The diffusion of ferric ions through these gels causes degradation with time of the dose distribution image. It would be useful to reduce ferric ion diffusion without decreasing gel sensitivity. The amount of ferric ion diffusion is a function of the time delay after radiation, the gel temperature, and the gel concentration. These effects can be quantified by measuring the ferric ion diffusion coefficient. Determination of the diffusion coefficient by irradiating the lower section of a cylinder of gel, which was then imaged repeatedly over time with a clinical magnetic resonanceimager, is described. Analysis of the edge spread function formed at each of several times after irradiation by drawing a profile over the imaged junction between the irradiated and unirradiated halves of the cylinder, gave estimates of the variance of the edge spread function. These variances were used to obtain an estimate of the ferric ion diffusion coefficient for the gel. A method of reducing ferric ion diffusion by adding a chelator and the cross linkage agent formaldehyde is suggested. The chelators investigated were 1,10 phenanthroline, xylenol orange, and bathophenanthroline disulfonic acid. These reduced diffusion to varying extents, and influenced the gel sensitivity. The diffusion coefficient in gels containing xylenol orange was found to be 0.44 mm2h−1. The gel sensitivity was 0.0093 s−1Gy−1. This compared with a diffusion coefficient of 0.82 mm2h−1 for the base line gel that did not contain formaldehyde or chelators. The sensitivity of this base line gel was 0.0129 s−1Gy−1. The addition of xylenol orange produced the most improved gel dosimeter of the gels studied. This gel had a decreased ferric ion diffusion coefficient and a decreased sensitivity. It was still sensitive enough to be useful.