To investigate whether objective polysomnographic measures of prevalent sleep problems such as sleep-disordered-breathing (SDB) and insomnia are associated with activities of daily living levels in inpatients at rehabilitation units.Retrospective and observational study.Single rehabilitation center.Inpatients with subacute stroke (N=123) (61.6±13.1 years; 23.8±3.4 kg/m2; 33% women; 90.5±36.7 days post-stroke) underwent a 1-night polysomnographic study and a 1-month inpatient rehabilitation program.Admission and discharge Barthel Index (BI) scores and its change scores.One hundred three (92%) patients had moderate-to-severe SDB (46.7±25.1 events/h in the apnea-hypopnea index), and 24 (19.5%) patients had acceptable continuous positive airway pressure adherence. Diverse values were found for total sleep time (259±71 min), sleep efficiency (69.5%±19.3%), sleep latency (24.3±30.9 min), and wakefulness after sleep onset (93.1±74.2 min). Admission BI scores and the BI change scores were 33.8±23.2 and 10.1±9.2, respectively. The National Institutes of Health Stroke Scale (NIHSS, 10.2±5.6), available in 57 (46%) patients, was negatively associated with admission levels and gains in BI change scores (P<.001, =0.002, respectively) in a univariate analysis. In regression models with backward selection, excluding NIHSS score, both age (P=.025) and wakefulness after sleep onset (P<.001) were negatively associated (adjusted R2=0.260) with admission BI scores. Comorbidity of hypertension; sleep latency percentage of stage 1, non–rapid eye movement sleep; and desaturation events ≥4% (P<.001, 0.001, 0.021, and 0.043, respectively; adjusted R2=0.252) were negatively associated with BI score gains.Based on objective sleep measures, insomnia rather than SDB in inpatients with subacute stroke was associated negatively with admission levels of activity of daily living and its improvement after a 1-month rehabilitation course.