OBJECTIVES: Investigation of the occurrence of postoperative hemorrhage after tonsillectomy and risk factors for these complications. STUDY DESIGN: Prospective National Audit with electronic web-based data collection from the National Health Service and independent hospitals in England and Northern Ireland. METHODS: Consenting patients undergoing tonsillectomy between July 2003 and September 2004 were included. A central database of patient and surgical variables was developed for analysis of primary and secondary postoperative hemorrhage and development of risk models for tonsillectomy complications. RESULTS: The Audit received data from 277 hospitals in England and Northern Ireland on 40,514 patients. Analysis was conducted on 33,921 (84%) who gave consent. Over the whole Audit 1,197 (3.5%) postoperative hemorrhages were recorded. One hundred eighty-eight (0.6%) patients sustained a primary hemorrhage and 1,033 (3%) a secondary hemorrhage (24 had both). Elevated hemorrhage rates were observed in tonsillectomies performed using diathermy for dissection and hemostasis compared with cold steel dissection and ties for hemostasis. National guidance issued midway through the Audit influenced practice and reduced hemorrhage rates. Risk models for postoperative hemorrhage were developed incorporating the effect of the guidance. For hemorrhage, compared with the cold steel group, bipolar diathermy tonsillectomy had an odds ratio of 2.47 (1.81-3.36), P < .0001, and bipolar diathermy scissors tonsillectomy an odds ratio of 3.20 (2.09-4.90), P < .0001. Use of bipolar diathermy for hemostasis only after cold steel dissection carried an intermediate risk, odds ratio 1.57 (1.16-2.13), P = .004. CONCLUSIONS: The results confirm that "hot" tonsillectomy techniques carry a substantially elevated risk of postoperative hemorrhage when diathermy is used as a dissection tool in tonsillectomy.