To assess the type of evidence that supported our management of surgical patients. Retrospective audit. Teaching hospital, Republic of Ireland. All 222 patients admitted by two surgical teams during one month. Patients admitted for diagnostic procedures were included if they were treated as a result. Diagnosis and treatment were established from the notes. Evidence for each intervention was then sought in MEDLINE 1986-1999, Best Evidence 1991-1999, and the Cochrane Database of Systematic Reviews Issue 3, 1999, and graded into one of five categories: systematic review, meta-analysis, randomised controlled trial, prospective study, or retrospective study. Treatment was supported by systematic review in 14, by meta-analysis in 12, by randomised controlled trial in 75, by prospective studies in 55, by retrospective studies in 58, and by no evidence in 8. Everyday surgical practice in our unit is supported by good quality evidence from recent publications.