It appears that uvulopalatopharyngoplasty (UVPP) is a reliable procedure for reducing snoring, but much less reliable when used as a treatment for OSAS. This is thought to be because of poor patient selection in that the site of the problem is not always the site of the operation. We present the technique of sleep nasendoscopy which allows direct visualization of the site or sites of obstruction in a sleeping patient. Our study has shown that there are patients with obstructive sleep apnoea syndrome (OSAS) in whom the only site of pharyngeal obstruction is at the velopharynx. These patients should do well with the relatively simple procedure of UVPP. This is not true for many other OSAS patients in whom we found that obstruction was multisegmental. This helps to explain the frequently poor results of UVPP in OSAS patients. We feel that this form of preoperative assessment will avoid unnecessary surgery.