Background: For cardiac arrest survivors initially comatose after restoration of spontaneous circulation (ROSC), the extent of brain injury and expected neurologic outcome are crucial for patient management decisions. Advanced diffusion imaging approaches such as neurite orientation dispersion and density imaging (NODDI) or diffusion kurtosis imaging may provide additional insight into tissue integrity and potential for recovery of consciousness complementary to standard diffusion tensor imaging (DTI). Methods: Multi-shell diffusion imaging was acquired in a prospective study of comatose cardiac arrest patients and in 5 controls. Neurite orientation dispersion (OD), intracellular volume fraction (ICVF), mean kurtosis (MK), axial kurtosis (AK), radial kurtosis (RK), mean diffusivity (MD), axial diffusivity (AD), radial diffusivity (RD) and fractional anisotropy (FA) were calculated. Median whole-brain values in patients with poor outcomes (no arousal recovery [AR] by discharge) were compared with those with AR and to controls (1-way ANOVA, post-hoc 1-sided Wilcoxon exact test). Results: 18 patients (mean ±SD 48±23 y, 39% men) and 5 controls (37±19 y, 40% men) were analyzed. Median (range) Glasgow Coma Scale was 3 [3-5]. 10 patients exhibited AR, 8 did not. Median [IQR] time-to-MRI was 5 [4-8] days. FA (P=0.009), MK (P=0.017), AK (P=0.026), RK (P=0.014), OD (P=0.018) and ICVF (P=0.0038) were significantly different (see Figure). FA control values were greater than AR and no AR (P<0.05). MK, AK, RK, OD and ICVF values in the no AR group were greater than in the AR and control groups (P<0.05). Discussion: This is the first report investigating early NODDI and diffusional heterogeneity changes in post-cardiac arrest comatose patients. Patients who failed to recover arousal demonstrated greater values for all kurtosis and NODDI metrics compared to controls. Potential bias from early withdrawal of life sustaining treatment and small cohorts are limitations.