Background: Esophageal pressure, used as a surrogate for pleural pressure, is commonly measured by air-filled balloon, and the accuracy of measurement depends on the proper balloon volume. It has been found that larger filling volume is required at higher surrounding pressure. In the present study, we determined the balloon pressure-volume relationship in a bench model simulating the pleural cavity during controlled ventilation. The aim was to confirm whether an optimal balloon volume range existed that could provide accurate measurement at both end-expiration and end-inspiration. Methods: We investigated three esophageal balloons with different dimensions and materials: Cooper, SmartCath-G, and Microtek catheters. The balloon was introduced into a glass chamber simulating the pleural cavity and volume-controlled ventilation was initiated. The ventilator was set to obtain respective chamber pressures of 5 and 20 cmH2O during end-expiratory and end-inspiratory occlusion. Balloon was progressively inflated, and balloon pressure and chamber pressure were measured. Balloon transmural pressure was defined as the difference between balloon and chamber pressure. The balloon pressure-volume curve was fitted by sigmoid regression, and the minimal and maximal balloon volume accurately reflecting the surrounding pressure was estimated using the lower and upper inflection point of the fitted sigmoid curve. Balloon volumes at end-expiratory and end-inspiratory occlusion were explored, and the balloon volume range that provided accurate measurement at both phases was defined as the optimal filling volume. Results: Sigmoid regression of the balloon pressure-volume curve was justified by the dimensionless variable fitting and residual distribution analysis. All balloon transmural pressures were within +/- 1.0 cmH(2)O at the minimal and maximal balloon volumes. The minimal and maximal balloon volumes during end-inspiratory occlusion were significantly larger than those during end-expiratory occlusion, except for the minimal volume in Cooper catheter. Mean (+/- standard deviation) of optimal filling volume both suitable for end-expiratory and end-inspiratory measurement ranged 0.7 +/- 0.0 to 1.7 +/- 0.2 ml in Cooper, 1.9 +/- 0.2 to 3.6 +/- 0.3 ml in SmartCath-G, and 2.2 +/- 0.2 to 4.6 +/- 0.1 ml in Microtek catheter. Conclusions: In each of the tested balloon, an optimal filling volume range was found that provided accurate measurement during both end-expiratory and end-inspiratory occlusion.