Diffusion Weighted Imaging (DWI) has provided a new window into brain connectivity reorganization and recovery following stroke^[1]. Besides, DWI has proven its utility in predicting patients’ clinical outcome^[2], using both parametric maps^[3] and tractography metrics^[4,5]. A previous work from our group evidenced that Generalized Fractional Anisotropy (GFA), as derived from DSI data in the uninjured cortical motor network, represents a strong predictor of stroke recovery^[4]. In this work, we aimed at investigating whether the structural properties of cortico-subcortical (CS) motor circuits, as measured by Diffusion Spectrum Imaging (DSI) metrics, are related to motor outcome after stroke. To do this, we focused on the CS motor circuits and investigated a number of recently introduced microstructural indices that may be derived from DSI: (i) the Propagator Anisotropy (PA), (ii) the Return to axis Probability (RTAP) and (iii) an estimation of the axon’s Radius (R), which are derived from the 3D Simple Harmonic Oscillator based Reconstruction and Estimation^[6] (3D-SHORE) model. We then used those indices in a linear regression model jointly with acute clinical scores, stroke size and age to assess their relationship with clinical outcome at six months after stroke. The target analysis focused on the uninjured cortico-subcortical (CS) motor-loop [primary motor areas (M1) - putamen (Put) - globus pallidus (GPi) - ventral lateral thalamic nucleus (Thal) - motor cortex (M1)] in ischemic stroke patients.Ten ischemic stroke patients with strokes affecting the motor cortex (age = 56.1 ± 17.8; female:male=4:6) were enrolled in the study. All subjects underwent 3 DSI scans (repetition time [TR]/echo time [TE]=6,600/138 msec, field of view [FOV]=212x212 mm, 34 slices, 2.2x2.2x3 mm resolution, 258 diffusion directions, b=8,000 s/mm²)within one week (tp1), one month (± one week, tp2), and six months (± fifteen days, tp3) after stroke. Patients underwent clinical assessments (NIH Stroke Scale [NIHSS] scores) at each time point. The motor part of the NIHSS score (NIHSS motor) was derived from items 2 (gaze) to 7 (ataxia) and 10 (dysarthria) (http://www.nihstrokescale.org/). The regions of interest in the CS motor loop were extracted from high resolution MPRAGE images (TR/TE=2,400/3 ms, voxel=1x1x1.2 mm, FOV=256x240 mm, generalized auto calibrating partially parallel acquisition [GRAPPA]=2) using Freesurfer (surfer.nmr.mgh.harvard.edu/) and manually corrected by an expert neurologist. The ODFs were reconstructed using Diffusion Toolkit (www.trackvis.org/dtk) and fiber tracking was performed via a streamline algorithm (www.cmtk.org). 3D-SHORE based indices were computed along fibers wiring the contralesional CS motor network, as illustrated in Fig. 1. Histograms of the indices were derived and some descriptors were considered (mean, variance, skewness, kurtosis). A linear regressor model applying backward selection^[4,7] was used to estimate the predictive power of each 3D-SHORE feature for the clinical outcome at 6 months (NIHSS3 motor). Each feature was processed independently to reduce the number of predictive variables in a small cohort of subjects.The mean value of RTAP along all CS motor loop connections at tp1, together with the stroke size at the same time point, was the variable that best correlated with NIHSS motor at tp3 (R²=0.96; R²adj=0.90,Table 1). Remarkably, RTAP represents the reciprocal of the mean cross-sectional area of the pore^[6] such that an estimation of the axonal radius can be inferred from it as $$$R=√(π/RTAP)$$$. These results point therefore at a possible relationship between axonal diameter in the acute stroke phase and the potential of functional recovery six months after the acute event. Other good predictive models were obtained combining the mean GFA, PA and R characteristics along the M1-Thal connection with the NIHSS motor at tp1 (R²=0.82±0.03; R²adj=0.77±0.05 Table 1). Mean values of 3D-SHORE indices were overall better correlated with motor outcome at 6 months after stroke than their variance, skewness and kurtosis. Interestingly, the microstructural characteristics of the connections GPi-Put and M1-Thal were among the best predictors of NIHSS motor at tp3. Our results show that RTAP histogram features across the CS motor loop provided the best prediction of motor outcome 6 months after stroke. Furthermore, our data confirm that the pre-stroke characteristics of thalamo-cortical connections are strongly related to stroke recovery at 6 months follow-up, as recently suggested^[8]. In addition, we provide new evidence of the importance of intrastriatal connectivity (GPi-Put) in stroke recovery, which warrants further investigations.