Synthetic magnetic resonance (MR) imaging refers to the procedure used to predict magnetic resonance images for any given design parameter settings using at least three observed MRI scans. SynMRI is an R packaged we developed for this purpose. We implemented the method proposed in Maitra and Riddles (2010), which used a model based on the Bloch equation, an empirical expression describing the nuclear magnetic resonance phenomena, to get voxel-wise estimates that are used to predict the intensity values for a given design parameters settings. The noise on the MR signal is modeled using the Rice distribution. All the parameters involved are estimated using the EM algorithm. SynMRI includes functions the compute the estimates, and visualize and evaluate the results. The EM algorithm estimation stage is performed in C since it is the part most compute intensive.