We measured nasal resistance (Rn) while awake, during midazolam-induced sedation and after antagonism with flumazenil (n = 9). Nasal and oral airflow were measured. Rn was calculated by dividing the difference between maximal nasal mask and oropharyngeal pressures by inspiratory airflow at minimum pharyngeal pressure. During sedation, two subjects developed obstructive apnoeic events and four subjects had snoring events. Each apnoea was ended by mechanisms other than a change in breathing route. After antagonism with flumazenil, apnoeic and snoring events were abolished. Rn during midazolam sedation (median 1.46 (25th percentile 1.00, 75th 2.61) kPa litre-1 s) was significantly greater than before midazolam (0.29 (0.25, 0.48) kPa litre-1 s) and after flumazenil (0.41 (0.25, 0.58) kPa litre-1 s) (P < 0.01 in each subject). We conclude that midazolam increased Rn, sometimes leading to obstruction, and flumazenil abolished this increase in Rn.