Validation of a computer model is described. The behavior of this model is compared both with mechanical ventilation of a test lung in a laboratory setup that uses a washout method and with manual ventilation. A comparison is also made with results obtained from a volunteer breathing spontaneously through a Bain circuit and with results published in the literature. This computer model is a multisegment representation of the Bain circuit and connecting tubing. For each segment, gas pressure, gas volume flow, and partial pressure of carbon dioxide are calculated for any number of breaths wanted. As a result, the time course of these variables can be generated for any location or, conversely, the carbon dioxide distribution in the system can be calculated for any time instant. A test lung, the human lungs, the ventilator bellows, and the reservoir bag are each represented by a single segment. The shapes of pressure and flow curves and of the capnograms taken at different locations in the Bain tubing are in good agreement. The washout study permits measurement of the time delay between the first expiration and the arrival of carbon dioxide at a particular location. The carbon dioxide level in the test lung decreases during inspiration and is stable during expiration. Quantitative agreement between model and experimental transport delays and carbon dioxide levels is such that the differences can be explained by the inaccuracy of the measurement. This is concluded from a sensitivity analysis. The study of the effect of segment size shows an almost optimal agreement between model behavior and experimental results for a 36-segment model. Execution of a thorough validation is imperative before such models can be used for clinical management and decision making or for teaching.