The possibility of reducing recovery time after colonoscopy was studied using patient-administered nitrous oxide and comparing it with our standard treatment of ketobemidone plus midazolam. Fifty consecutive colonoscopy patients were randomized to receive either (i) intravenous ketobemidone hydrochloride 2.5 mg, midazolam 2.5 mg, and breathing air from a face mask with a demand valve (KHM) or (ii) intravenous saline and a breathing mixture of even parts of oxygen and nitrous oxide (Entonox) from the same valve setup. Patient discomfort during colonoscopy was assessed using visual analogue scales. All patients were allowed to stay for recovery as long as they wanted, and the time was measured. Modified recollection tests were performed prior to colonoscopy and when the patients left the Endoscopy Unit, in order to study the degree of mental impairment induced by the procedure and the medication. All patients had complete colonoscopies of the same duration in both groups. Discomfort during colonoscopy was rated the same in both groups (2p = 0.6413). Both groups of patients scored identically in the precolonoscopy recollection test. Most patients had a lower score after colonoscopy, but Entonox-treated patients scored significantly better than those with KHM (2p = 0.0250). Patients treated with Entonox opted to leave the Unit directly after the procedure (median 0 minutes; interquartile range 0-5 minutes) compared to 38 minutes for those with KHM (interquartile range 10-75 min), 2p < 0.001. It seems from our data that nitrous oxide gives pain relief equal to that in our standard treatment.(ABSTRACT TRUNCATED AT 250 WORDS)