Objective: To comparing the effects of unilateral and bilateral mirror therapy on upper extremity function of stroke at early subacute stage. We hypothesize there are different effects between unilateral and bilateral mirror therapy on stroke patients at early subacute stage. Method: Patients with unilateral stroke and the onset within one month will be recruited and then randomly allocated to one of the three groups (including unilateral mirror therapy, bilateral mirror therapy, and conventional occupational therapy). Patient will receive 20 consecutive sessions of intervention (5 time per week, totally 4 weeks) and assessments before and after the intervention within one week. In each intervention session, patients will receive 30 minutes unilateral or bilateral mirror therapy depending on their allocated groups and then 30 conventional occupational therapy. The patients recruited in the conventional occupational therapy group will receive 60 minute conventional therapy. The outcome measures at pre- and post-treatment will including the Fugl-Meyer Assessment for upper extremity (FMA-UE), the Modified Ashworth Scale (MAS), the Jamar Hydraulic Hand Dynamometer, the Chedoke Arm and Hand Activity Inventory (CAHAI), and the Revised Nottingham Sensory Assessment (rNSA).

Inclusion Criteria:

• Unilateral stroke onset within 1 month
• The score of Mini-Mental State Examination (MMSE) is more than 24
• The score of Fugl-Meyer Assessment for upper extremity (FMA-UE) is between 18 to 56
• The score of Modified Ashworth Scale (MAS) is less than 3
• Willing to receive 3-5 sessions of the intervention per week, total 20 sessions consecutively
• Willing to sign informed consent

Exclusion Criteria:

• Unstable vital sign or complicating with other symptoms of neurological disease
• Auditory or visual function impairment
• Complicating with perceptual impairment (e.g., apraxia, neglect, or visual agnosia)
• Receiving botulinum toxin injection within 3 months
• Complicating with Wernicke's or Broca's aphasia

The outcomes assessors in charrge of pre- and post-test assessments were blinded to randomized allocation and intervention.