To assess whether ultrasound detection of vault haematoma can be used as a predictor of post-operative morbidity following vaginal hysterectomy. Prospective observational study of consecutive cases of vaginal hysterectomy performed between 1991 and 1994. Two hundred and twenty-three women having undergone vaginal hysterectomy. All women had transvaginal ultrasound examination by an independent observer on the third post-operative day. Routine observations and clinical assessments were made by established management protocol, by staff blinded to the ultrasound findings. Febrile morbidity; haemoglobin drop; need for blood transfusion; length of stay in hospital; re-admission rate and length of stay. Of the 223 scanned women, 55 (25%) had a vault haematoma. This group was compared with the 168 women with no haematoma. Significant increases in febrile morbidity (31% vs 7%), post-operative haemoglobin drop (2.5 g/dL vs 1.6 g/dL), need for blood transfusion (14.5% vs 1.2%), representation to hospital (25.5% vs 5.4%) and length of hospital stay (8.87 days vs 6.25 days) were seen in the haematoma group. Ultrasound detection of vault haematoma following vaginal hysterectomy is a common finding associated with increased febrile morbidity, need for blood transfusion, longer hospital stay and higher re-admission rate. In view of increasing demand for early discharge, driven by purchasers and patients, post-operative ultrasound identifies a high risk population and is both convenient and noninvasive. To further reduce morbidity, it also offers the opportunity to perform a controlled trial of prophylactic antibiotics in this identified subset.