Fractional anisotropy (FA) is an effective marker of motor outcome at the chronic stage of stroke yet proves to be less efficient at early time points. This study aims to determine which diffusion metric in which location is the best marker of long-term stroke outcome after thrombolysis with diffusion tensor imaging (DTI) at 24 hours post-stroke. Twenty-eight thrombolyzed patients underwent DTI at 24 hours post-stroke onset. Ipsilesional and contralesional FA, mean (MD), axial (AD), and radial (RD) diffusivities values were calculated in different Regions-of-Interest (ROIs): (1) the white matter underlying the precentral gyrus (M1), (2) the corona radiata (CoRad), (3) the posterior limb of the internal capsule (PLIC) and (4) the cerebral peduncles (CP). NIHSS scores were acquired at admission, day 1, and day 7; modified Rankin Scores (mRS) at 3 months. Significant decreases were found in FA, MD, and AD of the ipsilesional CoRad and M1. MD and AD were also significantly lower in the PLIC. The ratio of ipsi and contralesional AD of the CoRad (CoRad-rAD) was the strongest diffusion parameter correlated with motor NIHSS scores on day 7 and with the mRS at 3 months. A Receiver-Operator Curve analysis yielded a model for the CoRad-rAD to predict good outcome based on upper limb NIHSS motor scores and mRS with high specificity and sensitivity. FA values were not correlated with clinical outcome. In conclusion, axial diffusivity of the CoRad from clinical DTI at 24 hours post-stroke is the most appropriate diffusion metric for quantifying stroke damage to predict outcome, suggesting the importance of early axonal damage.