Fifty patients at risk for postoperative lung dysfunction and undergoing elective coronary revascularization have been randomly assigned to receive normothermic (36 degrees C) perfusion with warm heart protection (NP group) or hypothermic (28 degrees C) perfusion with cold heart protection (HP group). Lung function before and after cardiopulmonary bypass (CPB) was studied through the determination of the intrapulmonary shunt (Qs/Qt), the alveoloarterial oxygen gradient (A-a delta O2), and the artero-alveolar carbon dioxide gradient (a-A delta CO2). The Qs/Qt after CPB was significantly lower in the NP group (27.1 +/- 2.6 vs 35.7 +/- 2.3) as well as the A-a delta O2 (50.2 +/- 1.5 vs 57.6 +/- 2.4), both data returned to comparable between the groups after 3 h in the intensive care unit. The a-A delta CO2 was significantly lower after CPB in the NP group (5.2 +/- 0.74 vs 8.2 +/- 0.8). Hospital stay and mortality were comparable in the two groups; intubation time and rate of early extubation showed a trend in favour of the NP group; the rate of patients suffering hypoxia and/or hypercapnia after extubation was significantly lower in the NP group (12%) versus the HP group (44%). Normothermia seems to exert a protective effect against lung dysfunction after CPB. The absence of a rewarming injury associated with reperfusion, a limitation of the hypothermic-induced vasoconstriction due to local cooling of the lung and a better compliance of the normothermic lung are hypothesized as beneficial effects of the 'all-warm' strategy.