To systematically review sleep evaluation, characterize sleep disruption, and explore effects of sleepdisruption on outcomes in adult ICU patients. We systematically searched databases from May 1969 to June 2021 (PROSPERO protocol number: CRD42020175581). Prospective and retrospective studies were included studying sleep in critically ill adults, excluding patients with sleep or psychiatric disorders. Meta-regression methods were applied when feasible. 132 studies (8797 patients) were included. Fifteen sleep assessment methods were identified, with only two validated. Patients had significant sleep disruption, with low sleep time, and low proportion of restorative rapid eye movement (REM). Sedation was associated with higher sleep efficiency and sleep time. Surgical versus medical patients had lower sleep quality. Patients on ventilation had a higher amount of light sleep. Meta-regression only suggested an association between total sleep time and occurrence of delirium (p < 0.001, 15 studies, 519 patients). Scarce data precluded further analyses. Sleep characterized with polysomnography (PSG) correlated well with actigraphy and Richards Campbell Sleep Questionnaire (RCSQ). Sleep in critically ill patients is severely disturbed, and actigraphy and RCSQ seem reliable alternatives to PSG. Future studies should evaluate impact of sleep disruption on outcomes.