Partial paralysis of the hand is one of the main impairments after stroke. Constraint Induced Movement Therapy (CIT) is a new treatment technique that appears to improve upper extremity function after stroke. CIT consists of 6 hours of training/day for the affected arm (mainly with shaping exercises) and of restraint (mitt) of the non affected arm for two weeks. There are concerns about the practicality and resource issues in carrying out CIT according to the original model. In this thesis the benefit of modifications of CIT, of an assessment tool and of two common types of hand training have been evaluated.   CIT (n=16) administered in groups for two weeks (paper I) seems to be a feasible alternative to improve upper limb motor function after chronic stroke. The arm/hand motor performance improved significantly on Motor Assessment Scale (MAS; p= 0.003) and on Sollerman hand function test (p= 0.037). The median self reported motor ability (MAL) also improved (p < 0.001). No additional effect was seen from wearing a mitt for an extended period of three months. The reliability of the Sollerman hand function test (paper II) was studied in patients with chronic stroke. Three examiners observed 24 patients at three experimental sessions. There was agreement (kappa ≥ 0.4) between the examiners for 15/20 subtests. Using total sum scores, the agreement within the examiners was higher than 0.96 (for Spearman’s rhos and ICCs) and agreement between the examiners was higher than 0.96 (Spearman’s rhos) and 0.92 (ICCs), respectively. In a cohort of 24 patients with subacute stroke (paper III) forced use therapy (FUT; mitt use and 3 hours of training/day for 2 weeks) improved arm/hand function, but not more than regular arm therapy given to the control group. Significant improvements in arm/hand motor performance were found in the FUT group (n=12) as well as in the control group (n=12) on the Sollerman hand function test (p= 0.001), on MAS (p< 0.05) and on MAL (p < 0.05). No significant differences were seen between the groups pre- or post training or at three months follow up, demonstrating that the mitt had limited importance. In a separate study on 30 healthy subjects (paper IV), employing transcranial magnetic brain stimulation (TMS), we found that shaping exercises but not general activity training increased dexterity (p<0.05; Purdue peg board test) of the trained non dominant hand. After shaping exercises the cortical motor map shifted forwardly into the premotor area but did not expand. After general activity training the cortical motor map expanded significantly (p=0.03) in the posterior (sensory) direction. Shift of location of active TMS positions rather than their numbers might therefore be a critical factor for the interpretation of cortical plasticity.   In conclusion, the studies in this thesis have shown that less resource consuming modifications of CIT may be feasible to improve upper limb motor function after stroke. The type and amount of training for the more affected arm seems to be an important factor rather than the mitt use in itself. Shaping exercises, at least in healthy people, are effective in improving dexterity and the Sollerman hand function test reliable to evaluate arm/hand function after stroke.