The purpose of this study is to investigate two aspects of robotic therapy after stroke. One goal is to determine if early robotic rehabilitation of the upper limb (beginning 5-9 days post-stroke) is more effective than later robotic rehabilitation (beginning 21-25 days post-stroke). The other goal is to determine if higher intensity robotic rehabilitation (2 hours/day) is more effective than lower intensity robotic rehabilitation (1 hour/day).

Medically stable stroke subjects will be recruited in the first few days following their stroke. All participants will complete clinical and robotic assessments of neurologic function at 7 time points. Therapy will occur daily (Monday through Friday) for 20 days.

Study participants will be randomly assigned to 1) start robot therapy early or late after stroke and 2)receive one or two hours of robot therapy per treatment day for four weeks, or 3) control group that will receive the current standard of care.

The participant's chart will be reviewed for information about their stroke and related health effects and medical treatments. Assessment points to track progress will occur at 7, 18, 31,44,90,180, and 365 days after a stroke for all groups (give or take 2 days to account for weekends and holidays).

Standard clinical assessments of neurologic function will be done at each assessment point and include: cognition, arm strength, muscle tone, spasticity, reflexes, dexterity, visual acuity and fields, the Behavioural Inattention Test, and tests of arm movement (Fugl-Meyer Upper-Extremity, Box and Block Test, Chedoke-McMaster Stroke Assessment, and the Action Research Arm Test). These assessments can usually be done in about an hour. The assessment may be done over two sessions if needed due to fatigue or scheduling conflicts.

Robotic therapy will be conducted using the Kinesiological Instrument for Normal and Altered Reaching Movements (KINARM, Bkin Technologies, Kingston, ON). It will include several different tasks, each designed to train aspects of sensorimotor function of the proximal upper limb. Task performance will be monitored and difficulty will increase within and between sessions. The majority of the investigator's methods have been used previously to achieve equivalent or superior outcomes to standard rehabilitation.

Robotic assessment will measure elbow and shoulder range of motion, reaching for targets, the ability to mirror match the position of an arm with the other arm, and the ability to use both arms to hit away moving targets. The robotic assessment will take approximately 1.5 hours.

Robotic tasks include:

Visually guided reaching with assistance or resistance; Virtual Soccer; Shape Tracking; Whack-a-mole; Table Tennis Task; Ball on Bar Task; Proprioceptive Reaching; Hand Ball; Proprioceptive Shape Tracking.

Inclusion Criteria:

• Recent first stroke (ischemic or hemorrhagic)
• Upper extremity Fugl-Meyer score 15-45
• Modified Ashworth score of shoulder/elbow less than or equal to 2
• Able to follow task instructions
• Visual acuity better than 20/50 in both eyes
• Able to give consent
• Able to commit to follow-up

Exclusion Criteria:

• Prior stroke or significant neurologic problem (e.g. Multiple Sclerosis)
• Pre-existing musculoskeletal injury that will interfere with active therapy
• Pre-Stroke Modified Rankin Score > 2
• Clinical evidence of Unilateral Spatial Neglect on the Behavioural Inattention Test (BIT)
• Enrollment in a concurrent clinical intervention trial
• Major co-morbid or concurrent illness such that improvement is unlikely or completion of the protocol as specified is unlikely