Abstract: Background. A recent Cochrane Review showed that early robotic training of the upper limb in stroke survivors can be more effective than other interventions when improving activities of daily living involving the arm function is the aim of therapy. Objective. We tested for efficacy of the study a protocol which involved the use of the NeReBot therapy in partial substitution of standard upper limb rehabilitation in post–acute stroke patients. Methods. In this dose-matched, randomized controlled clinical trial, 34 hemiparetic participants with movement against gravity in shoulder, elbow, and wrist muscle groups were enrolled within 15 days of the onset of stroke. All participants received a total daily rehabilitation treatment for 120 minutes, 5 days per week for 5 weeks. The control group received standard therapy for the upper limb. The experimental group received standard therapy (65% of exercise time) associated with robotic training (35% of exercise time). Muscle tone (Modified Ashworth Scale), strength (Medical Research Council), and synergism (Fugl-Meyer motor scores) were measured at impairment level, whereas dexterity (Box and Block Test and Frenchay Arm Test) and activities of daily living (Functional Independence Measure) were measured at activity level. All assessments were performed at baseline, at the end of therapy (time T1), at 3 months (time T2), and at 7 months (time T3) after entry. All between-group analyses were tested using nonparametric test with Bonferroni’s adjustments for multiple testing. Results. No significant between-group differences were found with respect to demographic characteristics, motor, dexterity, and ADLs at baseline, postintervention (T1) and at follow-up (T2 and T3). Conclusions. The robot therapy by NeReBot did not lead to better outcomes compared with conventional inpatient rehabilitation.