Robust and accurate methods of measuring relaxation times on commercial scanners have recently been developed. Such methods overcome many machine and sequence dependent errors, but are still subject to errors arising from motion artifacts. This paper considers methods of decreasing the effect of motion artifacts in in vivo T1 and T2 measurements. Three conventional techniques for reducing such artifacts are assessed for their use with relaxation time measurement sequences. It is shown that these techniques do not adversely affect the accuracy of phantom T1 measurements made with the spin echo/inversion recovery sequence, and of T2 measurements made using the PHAPS sequence. For in vivo measurements of T1 it is also shown that these techniques reduce the standard deviation of T1 measurements in regions of interest and improve the reproducibility of such measurements overall. Two other techniques for reducing motion artifacts were found not to be compatible with the accurate measurement of relaxation times.