ObjectiveTo investigate the predictors of minimal clinically important changes on outcome measures after robot-assisted therapy (RT). DesignObservational cohort study. SettingOutpatient rehabilitation clinics. ParticipantsA cohort of outpatients with stroke (N=55). InterventionsPatients with stroke received RT for 90 to 105min/d, 5d/wk, for 4 weeks. Main Outcome MeasuresOutcome measures, including the Fugl-Meyer Assessment (FMA) and Motor Activity Log (MAL), were measured before and after the intervention. Potential predictors include age, sex, side of lesion, time since stroke onset, finger extension, Box and Block Test (BBT) score, and FMA distal score. ResultsStatistical analysis showed that the BBT score (odds ratio[OR]=1.06; P=.04) was a significant predictor of clinically important changes in the FMA. Being a woman (OR=3.9; P=.05) and BBT score (OR=1.07; P=.02) were the 2 significant predictors of clinically significant changes in the MAL amount of use subscale. The BBT score was the significant predictor of an increased probability of achieving clinically important changes in the MAL quality of movement subscale (OR=1.07; P=.02). The R2 values for the 3 logistic regression models were low (.114−.272). ConclusionsThe results revealed that patients with stroke who had greater manual dexterity measured by the BBT appear to have a higher probability of achieving clinically significant motor and functional outcomes after RT. Further studies are needed to evaluate other potential predictors to improve the models and validate the findings.