When ventilating an unintubated patient with a self-inflating bag, high peak inspiratory flow rates may result in high peak airway pressure with subsequent stomach inflation; this may occur frequently when rescuers without daily experience in bag-valve-mask ventilation need to perform advanced airway management. The purpose of this study was to assess the effects of a newly developed self-inflating bag (mouth-to-bag resuscitator; Ambu, Glostrup, Denmark) that limits peak inspiratory flow. A bench model simulating a patient with an unintubated airway was used, consisting of a face mask, manikin head, training lung (lung compliance, 100 ml/0.098 kPa (100 ml/cm H(2)O)); airway resistance, 0.39 kPa/l per second (4 cm H(2)O/l/s), oesophagus (LESP, 1.96 kPa (20 cm H(2)O)) and simulated stomach. Twenty nurses were randomised to ventilate the manikin for 1 min (respiratory rate: 12 per minute) with either a standard self-inflating bag or the mouth-to-bag resuscitator, which requires the rescuer to blow up a single-use balloon inside the self-inflating bag, which in turns displaces air towards the patient. When supplemental oxygen is added, ventilation with up to 100% oxygen may be obtained, since expired air is only used as the driving gas. The mouth-to-bag resuscitator therefore allows two instead of one hand sealing the mask on the patient's face. The volunteers were blinded to the experimental design of the model until completion of the experimental protocol. The mouth-to-bag resuscitator versus standard self-inflating bag resulted in significantly (P<0.05) higher mean+/-S.D. mask tidal volumes (1048+/-161 vs. 785+/-174 ml) and lung tidal volumes (911+/-148 vs. 678+/-157 ml), longer inspiratory times (1.7+/-0.4 vs. 1.4+/-0.4 s), but significantly lower peak inspiratory flow rates (50+/-9 vs. 62+/-13 l/min) and mask leakage (10+/-4 vs. 15+/-9%); peak inspiratory pressure (17+/-2 vs. 17+/-2 cm H(2)O) and stomach tidal volumes (16+/-30 vs. 18+/-35 ml) were comparable. In conclusion, employing the mouth-to-bag resuscitator during simulated ventilation of an unintubated patient in respiratory arrest significantly decreased inspiratory flow rate and improved lung tidal volumes, while decreasing mask leakage.