Lung inflation is the most important, and most difficult step in newborn resuscitation. A wide variety of devices are used to achieve lung inflation, but there are relatively few data to guide clinicians in their choice of device. Methods: We tested the ability of instructors and trained candidates on a newborn life support course to deliver initial inflation breaths to a test lung, using a pressure limited blow-off valve, a 240-ml self-inflating bag and a 500-ml self-inflating bag in sequence. Results: Use of a 240-ml self-inflating bag was associated with shorter initial inflations of 1.8 s mean (95% CI 1.60¿1.99 s), compared with 2.42 s (2.24¿2.61 s), 2.40 s (2.08¿2.71 s) for 500-ml self-inflating bags and ¿Tom Thumb¿ T piece, respectively. Delivery of breaths within a target pressure range of 30 ± 5 cm H2O was significantly better using a T piece than either self-inflating bag (proportion within target range 0.05 (95% CI 0¿0.11), 0.17 (95% CI 0.12¿0.23), 0.89 (95% CI 0.83¿0.94) for 240-ml and 500-ml self-inflating bags and ¿Tom Thumb¿ T piece, respectively. Excessive pressure delivery with both sizes of self-inflating bag was frequent. Conclusions: These data do not support use of 240-ml or 500-ml self-inflating bags for resuscitation of newborn term infants. A variable pressure T piece blow-off system may be the easiest device to use for newborn resuscitation and the most reliable at delivering desired pressures for set times.