The use of positive end-expiratory pressure (PEEP) or lung recruitment maneuvers (RM) to improve oxygenation in acute respiratory distress syndrome (ARDS) is used but it may reduce cardiac output (CO). Intermittent PEEP may avoid these complications. Our objective was to determine if variable PEEP compared with constant PEEP is capable of maintaining arterial oxygenation and minimizing hemodynamic alterations with or without RM. Eighteen dogs with ARDS induced by oleic acid were randomized into three equal groups: group 1, low variable PEEP; group 2, high variable PEEP, and group 3, RM + high variable PEEP. All groups were submitted to constant PEEP, followed by variable PEEP (PEEP was increased from 5 to 10 cmH(2)O in group 1, and from 5 to 18 cmH(2)O in the other two groups). PaO(2) was higher in group 3 (356.2 +/- 65.4 mmHg) than in group 1 (92.7 +/- 29.7 mmHg) and group 2 (228.5 +/- 72.4 mmHg), P < 0.05. PaO(2) was maintained during variable PEEP except in group 2 (318.5 +/- 82.9 at constant PEEP to 228.5 +/- 72.4 at variable PEEP). There was a reduction in CO in group 3 after RM (3.9 +/- 1.1 before to 2.7 +/- 0.5 L*min(-1)*(m(2))(-1) after; P < 0.05), but there was not any difference between constant and variable PEEP periods (2.7 +/- 0.5 and 2.4 +/- 0.7 L*min(-1)*(m(2))(-1); P > 0.05. Variable PEEP is able to maintain PaO(2) when performed in combination with RM in dogs with ARDS. After RM, CO was reduced and there was no relevant difference between the variable and constant PEEP periods.