Background: Rapid accurate assessment of metabolic derangements is crucial in the critically ill. We evaluated if arterial blood gas (ABG) samples transported through a pneumatic tube system (PTS) agreed with values transported by a human courier. Methods: In this prospective study of 50-paired ABG samples, the couriered reference ABG was compared with those transported by PTS. Agreement was summarised by the mean difference with 95% limits of agreement (LOA) and Lin's concordance correlation (pc). Results: The mean (+/- SD) time from sampling to analysis was 35.7 +/- 23.2 (courier) and 38.6 +/- 22.1 (PTS) minutes. Agreement was good between courier and PTS for pH, PaCO(2), bicarbonate, oxygen saturation and PaO(2) values (pc > 0.97). Although the mean difference in PaO(2) values between PTS and courier was small (-0.9 mm Hg) and the agreement was good, individual differences were clinically significant (95% LOA -40.8 to 39.0). For PaO(2) < 160 mm Hg, analysis of PTS samples yielded erroneously high PaO(2) values and vice versa for PaO(2) > 160 mm Hg compared to manual courier. This suggested exaggerated oxygen movement between the blood sample and air in the PTS. Conclusions: In this study, analysis of samples transported through the PTS resulted in clinically unacceptable PaO(2) values. Delay in transport and analysis of ABG samples should be avoided and samples transported manually if they cannot be assessed on-site.