Nasal mask ventilation has been shown to be effective, but outcomes do not always match expectations because of mouth leaks, patient-ventilator asynchrony, or decreased upper airway patency. These developments are detected when they lead ultimately to circuit leaks that lower the effectiveness of ventilation through pressure loss, poor inspiratory triggering, and prolonged inspiratory time. The quality of sleep is affected, and adverse effects and treatment intolerance may arise. A number of ways to detect leaks and their practical consequences are proposed in this article. We applied 310 leak-detection procedures to 177 patients who had disappointing clinical, gasometric, or polysomnographic outcomes of ventilation. The leak-detection procedures varied according to the type of ventilation and the supposed underlying pathophysiological mechanism. Significant leaks were detected in 132 patients (76%); therapeutic changes were then prescribed to optimize outcomes. We present a practical method to apply in patients with suboptimal ventilation outcomes. If leaks can be detected during treatment, the probable cause of treatment failure can sometimes be established and possible pathophysiological mechanisms better understood. With this knowledge, it may be possible to improve ventilation.