We attempt to measure the proper motions of two magnetars -the soft gamma-ray repeater SGR 1900+14 and the anomalous X-ray pulsar 1E 2259+586 -using two epochs of Chandra observations separated by ∼ 5 yr. We perform extensive tests using these data, archival data, and simulations to verify the accuracy of our measurements and understand their limitations. We find 90% upper limits on the proper motions of 54 mas yr −1 (SGR 1900+14) and 65 mas yr −1 (1E 2259+586), with the limits largely determined by the accuracy with which we could register the two epochs of data and by the inherent uncertainties on two-point proper motions. We translate the proper motions limits into limits on the transverse velocity using distances, and find v ⊥ < 1300 km s −1 (SGR 1900+14, for a distance of 5 kpc) and v ⊥ < 930 km s −1 (1E 2259+586, for a distance of 3 kpc) at 90% confidence; the range of possible distances for these objects makes a wide range of velocities possible, but it seems that the magnetars do not have uniformly high space velocities of > 3000 km s −1 . Unfortunately, our proper motions also cannot significantly constrain the previously proposed origins of these objects in nearby supernova remnants or star clusters, limited as much by our ignorance of ages as by our proper motions.