Oxygen therapy is a mainstay of critical care medicine, yet its optimal therapeutic use has not been systematically evaluated. A detailed understanding of current practice in oxygen therapy in intensive care is required to enable future interventional studies. We aimed to describe current oxygenation practice in patients requiring ≥48 hours of mechanical ventilation (MV) at an academic tertiary referral centre. We collected longitudinal arterial blood gas and hourly oxygenation data from intensive care unit charts in a consecutive cohort of 40 trauma, 41 medical and 20 surgical patients for their first seven MV days, analysed data for 14,063 MV hours, and derived time-weighted averages (TWA) of variables for each 24-hour interval on MV for all patients.The TWA-FiO2 was 0.42 (95% CI 0.41 to 0.44) and TWA-SpO2 was 97.1% (95% CI 96.8 to 97.4) for the first seven MV days. TWA-PaO2 was >80 mmHg on 80% of MV days. TWA-FiO2 of ≥0.35 was used to achieve TWA-SpO2 >95% on 61% of MV days. Of 58 MV days with TWA-FiO2 ≥0.60, TWA-SpO2 ≥96% occurred on 28 (48%) days. Mean SpO2 and PaO2 in patients with severe acute lung injury (ALI) scores were higher than recommended targets. Wide variability in the mean SpO2 and PaO2 was observed in patients with comparable ALI scores. Inspired oxygen therapy in these MV patients was 'liberal', with PaO2 and SpO2 values generally above 80 mmHg and 96% respectively. An interventional study comparing current practice to more conservative targets (PaO2≍60 to 65 mmHg and/or SpO2≍90 to 92%) appears possible.