We recently demonstrated that desipramine reduces the sleep-related loss of upper airway dilator muscle activity and reduces pharyngeal collapsibility in healthy humans without obstructive sleep apnoea (OSA). The aim of the present physiological study was to determine the effects of desipramine on upper airway collapsibility and apnoea–hypopnea index (AHI) in OSA patients.A placebo-controlled, double-blind, randomised crossover trial in 14 OSA patients was performed. Participants received treatment or placebo in randomised order before sleep. Pharyngeal collapsibility (critical collapsing pressure of the upper airway (P_crit)) and ventilation under both passive (V′_0,passive) and active (V′_0,active) upper airway muscle conditions were evaluated with continuous positive airway pressure (CPAP) manipulation. AHI was quantified off CPAP.Desipramine reduced activeP_crit(median (interquartile range) −5.2 (4.3) cmH₂O on desipramineversus−1.9 (2.7) cmH₂O on placebo; p=0.049) but not passiveP_crit(−2.2 (3.4)versus−0.7 (2.1) cmH₂O; p=0.135). A greater reduction in AHI occurred in those with minimal muscle compensation (defined asV′_0,active−V′_0,passive) on placebo (r=0.71, p=0.009). The reduction in AHI was driven by the improvement in muscle compensation (r=0.72, p=0.009).In OSA patients, noradrenergic stimulation with desipramine improves pharyngeal collapsibility and may be an effective treatment in patients with minimal upper airway muscle compensation.