During one-lung ventilation (OLV), titrating the positive end-expiratory pressure (PEEP) to target a low driving pressure (∆P) could reduce postoperative pulmonary complications. However, it is unclear how to conduct PEEP titration: by stepwise increase starting from zero PEEP (PEEP INCREMENTAL) or by stepwise decrease after a lung recruiting manoeuvre (PEEP DECREMENTAL). In this randomized trial, we compared the physiological effects of these two PEEP titration strategies on respiratory mechanics, ventilation/perfusion mismatch and gas exchange. Patients undergoing video-assisted thoracoscopic surgery in OLV were randomly assigned to a PEEP INCREMENTAL or PEEP DECREMENTAL strategy to match the lowest ∆P. In the PEEP INCREMENTAL group, PEEP was stepwise titrated from ZEEP up to 16 cm H 2 O, whereas in the PEEP DECREMENTAL group PEEP was decrementally titrated, starting from 16 cm H 2 O, immediately after a lung recruiting manoeuvre. Respiratory mechanics, ventilation/perfusion mismatch and blood gas analyses were recorded at baseline, after PEEP titration and at the end of surgery. Sixty patients were included in the study. After PEEP titration, shunt decreased similarly in both groups, from 50 [39-55]% to 35 [28-42]% in the PEEP INCREMENTAL and from 45 [37-58]% to 33 [25-45]% in the PEEP DECREMENTAL group (both p < 0.001 vs baseline). The resulting ∆P, however, was lower in the PEEP DECREMENTAL than in the PEEP INCREMENTAL group (8 [7-11] vs 10 [9-11] cm H 2 O; p = 0.03). In the PEEP DECREMENTAL group the PaO 2 / FIO 2 ratio increased significantly after intervention (from 140 [99-176] to 186 [152-243], p < 0.001). Both the PEEP INCREMENTAL and the PEEP DECREMENTAL strategies were able to decrease intraoperative shunt, but only PEEP DECREMENTAL improved oxygenation and lowered intraoperative ΔP. Clinical trial number NCT03635281; August 2018; "retrospectively registered"