OBJECTIVE: To evaluate and compare the effectiveness and tolerability of preoxygenation with the self-inflating bag-valve-mask (BVM) and non-rebreather mask (NRM) as are used before emergency anaesthesia. DESIGN: Device performance evaluation. SETTING: Experimental study. PARTICIPANTS: 12 male and 12 female healthy volunteers (age range 24-47) with no history of clinically significant respiratory disease. INTERVENTIONS: End-expiration oxygen measurements (F(E)O(2)) after 3 min of preoxygenation with BVM (without mechanical assistance) and NRM devices. Mask pressures were measured and subjective difficulty of breathing was also assessed with a visual analogue score (VAS). PRIMARY AND SECONDARY OUTCOME MEASURES: The final F(E)O(2) achieved was 58.0% (SD 7.3%) for the NRM compared to 53.1% (SD 13.4%) for the BVM (p=0.072). Preoxygenation was associated with small increases in F(E)CO(2) that were greater for the BVM (0.50%; 95% CI 0.48 to 0.52) than the NRM (0.29%; 95% CI 0.31 to 0.28); this difference was statistically significant (p=0.028). Both devices were well tolerated on VAS assessment of difficulty of breathing although this was higher for the BVM than the NRM (median VAS 1.85/10 compared to 1.1/10; p=0.041). Inspiratory and expiratory mask pressures were higher for the BVM. CONCLUSIONS: In healthy volunteers, the NRM performs comparably to the BVM in terms of the degree of denitrogenation achieved although neither performed well. Although it was well tolerated, the BVM was subjectively more difficult to breathe through and was associated with greater mask pressures and a small increase in F(E)CO(2) consistent with hypoventilation or rebreathing. Our results suggest that preoxygenation with the NRM may be a preferable approach in spontaneously breathing patients.