Functional connectivity studies commonly use resting-state BOLD MR images to study the neurodevelopment of healthy and at-risk neonates. BOLD images are highly sensitive to motion; post-acquisition motion correction techniques can be applied to BOLD data to compensate for motion. We compare the corrective performance of two motion correction techniques on a cohort of 17 healthy neonates: the traditional correction to the first volume technique and a novel, HMM-based motion correction technique. We evaluate the corrected images in terms of the Power et al. thresholds and show the HMM-based technique can be used to recover neonatal BOLD data corrupted by motion.