We hypothesised that hypocretin (orexin) plays a role in the determination of ventilatory chemosensitivity. 130 patients with narcolepsy-cataplexy (mean ± SD age 20 ± 10 yrs, 69% male) and 117 controls (22 ± 6.9 yrs, 62% male) were recruited and tested for human leukocyte antigen (HLA)-DQB1*0602 status, hyperoxia hypercapnic (change in minute ventilation (δV'(E))/carbon dioxide tension (δP(CO(2))) L·min(-1)·mmHg(-1)) and hypoxic (δV'(E) /change in arterial oxygen saturation measured by probe oximetry (δS(p,O(2))) L·min(-1) per %S(p,O(2))) responsiveness, and by spirometry. Hypocretin deficiency was determined either by measures of cerebrospinal fluid hypocretin-1 (37 patients) or by positive HLA-DQB1*0602 status. All patients and 49% of controls underwent polysomnography and multiple sleep latency testing. Despite similar spirometric values, patients had a higher apnoea/hypopnoea index (AHI) (2.8 ± 5.4 versus 0.8 ± 1.6 h(-1); p = 0.03) and lower minimal oxygen saturation during sleep (87% ± 7 versus 91 ± 4%; p = 0.0002), independent of age, sex and body mass index. Patients had depressed hypoxic responsiveness (0.13 ± 0.09 versus 0.19 ± 0.13 L·min(-1) per %S(p,O(2)); p<0.0001), independent of AHI, but hypercapnic responsiveness did not differ. Examined by HLA status, positive (26 out of 117) controls had lower hypoxic but similar hypercapnic responsiveness than those marker-negative (0.13 ± 0.08 versus 0.20 ± 0.14 L·min(-1) per %S(p,O(2)); p<0.0001). Thus, a lower hypoxic responsiveness in the narcolepsy-cataplexy group is a result of DQB1*0602 status rather than the clinical features of disease.