The incidence of stroke has not changed significantly in recent decades. However, the actual number of strokes is likely to increase as a consequence of the increasing number of elderly people, creating a significant burden on the health care system, the patients and their families. Pareses are the most common impairment reported after stroke. Motor impairments are associated with reduced self-perceived health, and consequently rehabilitation after stroke has a strong emphasis on physiotherapy and motor training. However, more knowledge about motor recovery and effects of therapy is needed for further improvement of rehabilitation processes and of outcome for the large number of stroke victims. The overall aim of this thesis was to increase the knowledge about motor recovery after stroke by evaluating the effect of two different rehabilitation programmes compared to standard rehabilitation regimes and by investigating changes in brain activity in patients treated in terms of the recommended guidelines in the acute phase and during follow-up. A randomised controlled trial was performed to evaluate the effect of an early supported discharge (ESD) service for patients with acute stroke living in a rural community. Sixty-two eligible patients were included and randomised to either an ESD service or to standard followup. The trial revealed no significant benefit on the Modified Rankin Scale, Barthel Index, Berg Balance Scale or walking speed, but significantly less isolation in the ESD group at the six-week follow-up. However, analysis of all cases with all assessments available showed a non-significant trend toward greater improvement in balance in the ESD group from one week to six weeks follow-up. The trial identified a strong association between initial severe leg paresis, but not with initial moderate leg paresis, and reduced balance one year after the stroke. There was also a strong association between initial inability to walk and reduced balance one year after the stroke. Another randomised controlled trial was performed to evaluate the efficacy and feasibility of Constraint-Induced Movement Therapy (CIMT) organised as group therapy for patients with subacute and chronic stroke. Thirty eligible patients were included and randomised to a CIMT group receiving ten days of intensive motor training of the affected arm or to a control group receiving standard rehabilitation. The CIMT group showed a statistically significant greater improvement in motor function of the affected arm at the post-treatment assessment. However, this difference did not persist at six months' follow-up. There were no differences between the groups at any time in relation to the amount of use of the affected arm or to independence in activities of daily living. Finally, a longitudinal follow-up study was performed to investigate the changes in brain activation patterns from the acute to the chronic phases and their relationship to motor learning after stroke. Twelve eligible patients with acute ischaemic stroke were included and assessed with functional magnetic resonance imaging (fMRI) and clinical tests within one week after stroke and three months later. All patients, except one, had complete recovery of the affected arm according to our criteria. Increased activation in cerebellum, striatum, angular gyrus and insula was revealed in the acute phase compared to the chronic phase. The chronic phase demonstrated a restoration of the lateralised primary motor network, in addition to increased bilateral somatosensory association areas and contralesional secondary somatosensory areas (SII). The activation patterns are not identical but comparable to a motor learning process. In conclusion, this thesis shows that the ESD service did not significantly influence death or dependency, balance or walking speed for patients living in a rural community. However, it may lead to less isolation and a transient improvement in self-perceived health. The thesis also demonstrates that CIMT organised as group therapy is feasible and efficient in the short term but may not be superior to standard rehabilitation in the long term. Finally, the thesis has revealed that the motor network changes associated with successful motor recovery are comparable to changes observed in motor learning studies in healthy subjects, and also indicating the importance of bi- and contralesional brain activation for successful motor recovery after stroke. ; PhD i klinisk medisin