Structural and functional MRI plays crucial roles in investigating subtle changes in the neonatal brain. Of the structural MRI techniques, susceptibility-weighted imaging (SWI) usually evaluates microhemorrhages and diffusional kurtosis imaging (DKI) can estimate microstructural alterations of axonal pathways and myelination. Functional MRI (fMRI), on the other hand, allows us to study the functional connectivity of the human cerebral cortex, typically measuring blood oxygen level-dependent (BOLD) signals with task-based paradigms. Recently, fMRI without tasks, so-called resting state fMRI (rs-fMRI), has become an important tool for evaluating functional organization in neonates. During MR scans, infants often require anesthesia because of frequent motion. However, this procedure increases the risk of adverse effects such as respiratory depression, especially in unstable neonates cared for in a neonatal intensive care unit. To avoid motion artifacts, several holding systems have been reported. Conventional devices may stabilize the head less effectively because of difficulty in keeping the head upward due to its dolichocephalic shape. We developed a new prototype body holder for MRI in unanesthetized neonates and evaluated its safety, usability, and image quality. We examined nine low-birth-weight neonates scanned between five days and two months of postnatal age (two with anesthesia using triclofos sodium syrup and seven without anesthesia using the holder). The custom-made device consists of two components: the holder based on an infant spinal immobilizer to fix the head, neck and trunk and a dedicated hat made of a prosthetic liner to fix the head (Fig. 1). The holding procedures were as follows: 1) the unanesthetized infant was placed in the holder, 2) the head was fitted to the hat, and then the jaw was firmly fixed to the holder, 3) the belt of the holder was fixed around the trunk and the infant was wrapped snugly with a towel (Fig. 2). This holder can keep the head straight and parallel to the body axis, which is typically required for fMRI, despite dolichocephaly. These procedures were easily performed by two staff outside the scanner room and required approximately five minutes. Immediately after setting the device, the infant was transferred to the scanner room. Structural images were routinely acquired using a 3.0T MR scanner for approximately 15 minutes. Axial images included turbo spin echo T1- and T2-weighted (T1WI and T2WI) and SWI, while sagittal images were acquired with T1-weighted, magnetization-prepared rapid gradient echo (MPRAGE) sequences. In addition to the four sequences, infants underwent DKI and rs-fMRI during the following 15 minutes, if consistently stable. Head motion was estimated by MCFLIRT in FSL during preprocessing for fMRI. We analyzed images qualitatively and quantitatively, comparing unanesthetized with anesthetized infants; qualitative analysis included diagnostic quality and noise/artifacts assessed by three radiologists independently on five-point scales, whereas quantitative evaluation consisted of contrast-to-noise ratio (CNR), comparing gray matter (GM), white matter (WM), putamen, medulla, pons, and cerebellum with cerebrospinal fluid (CSF).All the MR examinations were performed stably and uneventfully within 30 minutes for all infants (Fig. 3). Compared to the control score, average scores of the diagnostic quality for axial T2WI (1 min), T1WI (3 min), SWI (4.5 min) and sagittal T1WI (4.6 min) in the unanesthetized infants were 4.89±0.32, 4.61±0.61, 4.50±1.20 and 2.56±1.72, respectively. The average score of the noise/artifacts in the unanesthetized infants was 4.44±0.91. Especially, axial T2WI and T1WI of all the unanesthetized cases were evaluated to be as good as the anesthetized cases. Sagittal T1WI indicated a lower score because of motion artifacts caused by a pacifier in the mouth, although there were no artifacts extending to the intracranial regions. Average scores in CNR of GM/CSF, WM/CSF, putamen/CSF, medulla/CSF, pons/CSF and cerebellum/CSF on axial T1WI with anesthesia were 97.42±26.76, 77.93±5.32, 174.76±3.54, 165.09±2.72, 87.71±5.58 and 91.79±7.06. Without anesthesia they were 85.62±32, 60.88±28.69, 154.05±25.60, 146.99±37.40, 84.20±19.10 and 109.27±31.01. CNR of all sequences showed no significant differences between with and without anesthesia (p>0.05). Figure 4 shows motion estimations with and without anesthesia. The device was able to stabilize the neonate without anesthesia, whereas the anesthetized infant moved frequently despite anesthesia. The new holding device for infants provides the following benefits: First, major sequences are completed safely and relatively quickly. Second, the device is easily handled by two people with a short setup time. Finally, structural images can be obtained during natural sleep with reliable image quality. Additionally, rs-fMRI with the device has the potential to evaluate neural activation even in high-risk infants without anesthesia.The body holder can reduce the motion of neonates safely and improve imaging quality.