Objective: To show the effectiveness of goal-oriented proprioceptive training in subacute stroke for balance, autonomy, and fall risk. Methods: Out a total of 35 patients, consistent in age (75.31±8.65 years), type of stroke (ischemic, 3 to 11 weeks before), and motor impairment, 18 patients underwent solely proprioceptive rehabilitation, the other 17 dual task exercises. The study assessed autonomy using Barthel Index, fall risk with Timed Up and Go Test (TUG), balance through Berg Balance Scale (BBS) and Tinetti test. Results: After two months, significant improvements were recorded in Barthel Index, BBS (p<0.0001), Tinetti test (p<0.0001 in dual task group, p=0.0029 in single task group), and TUG (p=0.0052 in dual task group, p=0.0020 in single task group) in both groups. Comparing the two groups, dual task group showed a significant difference in Tinetti balance assessment (p=0.0052), between the total score of Tinetti test and TUG in single (p=0.0271), and dual task (p=0.0235). Likewise, Tinetti gait test was significantly related to TUG in single (p=0.0536), and dual task (p=0.0466), while Tinetti balance test to Barthel Index (p=0.0394), BBS (p<0.0001), and TUG in single (p=0.0219), and dual task (p=0.0196). Lastly, there is a positive correlation of the use of aids with BBS (p=0.0074), and total score of Tinetti test (p=0.0160). Conclusion In subacute stroke, goal-oriented proprioceptive training improved balance, but only partially autonomy. Furthermore, the use of aids after dual-task exercises improved recovery of balance, but did not reduced falls.