Our objective was to investigate correlations between clinical motor scores and cerebral sensorimotor activation to demonstrate that this reorganization is the neural substratum of motor recovery. Correlation analyses identified reorganization processes shared by all patients. Nine patients with first-time corticospinal tract lacuna were clinically evaluated using the NIH stroke scale, the motricity index, and the Barthel index. Patients were strictly selected for pure motor deficits. They underwent a first fMRI session (E1) 11 days after stroke, and then a second (E2) 4 weeks later. The task used was a calibrated repetitive passive flexion/extension of the paretic wrist. The control task was rest. Six healthy subjects followed the same protocol. Patients were also clinically evaluated 4 and 12 months after stroke. All patients improved significantly between E1 and E2. For E1 and E2, the ipsilesional primary sensorimotor and premotor cortex, supplementary motor area (SMA), and bilateral Broadmann area (BA) 40 were activated. Activation intensity was greater at the second examination except in the ipsilesional superior BA 40. Magnitude of activation was lower than that of controls except for well-recovered patients. E1 clinical hand motor score and E1 cerebral activation correlated in the SMA proper and inferior ipsilesional BA 40. Thus, we demonstrated early functionality of the sensorimotor system. The whole sensorimotor network activation correlated with motor status at E2, indicating a recovery of its function when activated. Moreover, the activation pattern in the acute phase (E1) had a predictive value: early recruitment and high activation of the SMA and inferior BA 40 were correlated with a faster or better motor recovery. On the contrary, activation of the contralesional hemisphere (prefrontal cortex and BA 39-40) and of the posterior cingulate/precuneus (BA 7-31) predicted a slower recovery.