Objective This systematic review aimed to examine the effects of home-based exercises in comparison with centre-based exercises for improving the paretic upper limb after stroke. Data sources AMED, MEDLINE, EMBASE CINAHL, Cochrane, PsycINFO, and PEDro databases. Review methods Only randomized clinical trials were included. Participants in the reviewed studies were adults at any time after stroke. The experimental intervention was home-based exercises compared with centre-based exercises. Outcome data related to strength, motor recovery, dexterity, activity, and participation were extracted from the eligible trials and combined in meta-analyses. The quality of included trials was assessed by the PEDro scores. The quality of evidence was determined according to the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) system. Results Eight trials, involving 488 participants, were included. Most trials (63%) delivered semi-supervised interventions (amount of supervision 3–43%), and three trials provided full supervision. Random-effects meta-analyses provided moderate- to high-quality evidence that home- and centre-based exercises provide similar effects on motor recovery (MD 1.4 points; 95% CI −0.9 to 3.8), dexterity (MD −0.01 pegs/s; 95% CI −0.04 to 0.05), upper limb activity performance (SMD −0.04; 95% CI −0.25 to 0.18), and quality of movement (0.1 points; 95% CI −0.2 to 0.4). Effects on strength were also similar but the quality of the evidence was rated as low. No trials examined effects on participation. Conclusion Effects of home-based prescribed exercises on upper limb motor recovery, dexterity, and activity are likely to be similar to improvements obtained by centre-based exercises after stroke.