BACKGROUND: Rehabilitation is a non-pharmacologic therapy that can restore health and reverse the patient's disability, but the efficacy of rehabilitation in critically ill patients is not well documented. METHODS: In a prospective cohort study, we assessed whether the degree of change in functional status after comprehensive rehabilitation influenced clinical outcomes in 77 tracheotomized patients (mean ± SD age 75 ± 7 y) admitted for difficult weaning to our regional weaning center. The care plan, including peripheral muscle training, was delivered daily. We recorded admission demographic, anthropometric, and functional characteristics. We measured the change in basic activities of daily living score (ΔBADL), survival, and weaning success rate as clinical outcomes. We performed the Pearson correlation analysis and linear regression, with ΔBADL as the dependent variable, to test the predictive power of the baseline measurements. RESULTS: Sixty-seven patients (87%) survived, and 55 of them (74%) succeeded in weaning during their stay in the weaning center. The mean ± SD ΔBADL improvement was 2.5 ± 2.0 points (median 2 points). Baseline performance of the latissimus dorsi predicted ΔBADL ( = 0.388, 95% CI 0.111‐1,664, P = .03). The probability of remaining ventilator-free (P = .043) and survival (P = .001) differed across the 4 ΔBADL categories (0 = no change, 1‐2 = least improvement, and > 2 = improvement above median change). CONCLUSIONS: Mortality rate and weaning success differ according to ΔBADL following active rehabilitation/training in tracheotomized, ventilated, difficult-to-wean patients. The performance of the latissimus dorsi was the only significant predictor of change.