For adults (mean age 61 years; 64% men) admitted to the ICU when specified due to traumatic brain injury, septic shock, chronic obstructive pulmonary disease, surgery, cardiac arrest, or stroke, evidence suggests that increasing oxygenation levels by at least 1 kPa in partial pressure of oxygen (PaO2), 10% in fraction of inspired oxygen (FiO2), or 2% in arterial oxygen saturation (SaO2)/pulse oximetry (SpO2) may increase mortality (347 vs 295 per 1000 people; all results on average) and the incidence of serious adverse events (SAEs) (333 vs 295 per 1000 people when the most common SAE reported in each trial was used) at six months follow‐up, and the incidence of sepsis within six days (94 vs 50 per 1000 people). However, adjustment for multiple outcomes, sparse data and repetitive testing (for mortality and SAEs), and the type of SAE data used in the analysis (event with highest proportion or cumulated event rate), impact results sufficiently to reduce confidence, downgrade the evidence to very low certainty, and preclude the possibility of drawing conclusions. When reported, the lower oxygenation levels used in trials were as follows: FiO2 0.40 to 0.50, PaO2 6.6 to 15 kPa, 50 to 112.5 mmHg, and/or SaO2/SpO2 88% to 98%.