Polysomnography is currently used for diagnosis, evaluation and selection of therapy in patients with obstructive sleep apnoea (OSA), but clinically successful uvulopalatopharyngoplasty (UPPP) is not necessarily reflected by post-operative improvement of polysomnographic recordings. Post-operative polysomnography may suggest deterioration of pre-existing OSA or, in snorers, de-novo precipitation of OSA. Thus, if polysomnography is a reliable indicator of OSA, then OSA may be a post-operative risk of UPPP. The aims of our study were: (i) to assess the possible deleterious effect of UPPP on sleep patterns; (ii) to further define the role of cardioisotope scanning in the evaluation of OSA; (iii) to assess the reliability of polysomnography given the clinical and cardioisotope scan findings. Symptoms, polysomnography and radionuclide ventriculography were prospectively compared pre- and post-operatively in 41 patients undergoing UPPP. In 12 patients (29 per cent), there were disparate results between pre- or post-operative polysomnography and the clinical and/or radionuclide ventriculography, as follows: In four of 16 patients with abnormal pre-operative ventricular performance, there was pre-operative symptomatology of severe OSA and a bedmate's reports of apnoeic episodes. This was in contrast to normal or near normal sleep apnoea recordings. In eight patients, post-operative improvement of symptoms was reported, despite deterioration of post-operative polysomnographic recordings. In these patients the post-operative improvement of symptoms was also reflected by improved ventricular performance. Worsening of ventricular performance was not demonstrated in any patient. In conclusion, UPPP does not induce OSA. Polysomnography may underestimate or even misdiagnose cases of OSA. The diagnostic importance of patient symptomatology should be stressed particularly in those patients with only mildly abnormal or even completely normal sleep studies. A combination of polysomnographic and cardiovascular evaluation in patients with symptomatology consistent with OSA is recommended.