The act of supplying air to the lungs is called ventilation. When the patient is unable to perform ventilation on his own, whether due to illness, birth defects, or anesthesia, the use of mechanical ventilation is recommended. In recent years, the pandemic caused by the coronavirus -Coronavirus Disease-2019 (COVID-19) has made evident the limitations of the Brazilian health system (SUS). Among them was the lack of lung ventilators. Most of the maintenance required for lung ventilators within hospitals is carried out by the Clinical Engineering Sector (CES). In addition, the CES must carry out scheduled services such as: preventive maintenance, electrical safety tests and calibration tests, in order to promote the reduction of risks to patients. Among these tasks, one of the most important is calibration. Its purpose is to evaluate the actual values provided by the ventilator in the respiratory cycle and verify if the equipment is delivering the correct amount of the selected parameter to the patient, in addition to making it possible to monitor changes in operation over its lifetime. Before a pulmonary ventilator can be commercialized in Brazil, several standards are applicable to obtain its registration for commercialization by ANVISA, one of which is the ABNT NBR ISO 80601-2-12 standard. It specifies performance tests in volume-controlled and pressure-controlled ventilation modes. However, after registration and commercialization, there is no legislation that guarantees the metrological reliability of the equipment. With this problem in mind, this work aimed to develop and test a computational system for evaluating the performance of pulmonary ventilators, after corrective maintenance, based on the ABNT NBR ISO 80601-2-12 standard and the use of the Guide for Expression of Uncertainty of Measurement for estimating uncertainties. A program based on LabView (National Instruments) software was developed to be used in conjunction with the Fluke Biomedical® VT650 gas analyser. The Michigan Instruments® TTL3 lung test was also used. Twenty-four lung ventilators, of 4 different models, which were in use at the hospital, were tested with the help of the computational program. The tests performed were based on what is stipulated by the first eight tests present in the ABNT standard, for volume-controlled and pressure-controlled ventilation modes, for an adult lung. With the help of the program, each ventilator can be tested in both ventilation modes in just one hour. As a result of the tests, none of the 24 deviceswere approved in all tests. It was also noted that the ventilators had a worse performance when in pressure-controlled mode. Other parameters that also obtained many disapprovals, in addition to peak inspiratory pressure, were the fraction of oxygen and the inspiratory time. In addition, the ventilators with fewer hours worked in use performed very close to the oldest ventilator in use in the hospital.