The Oxford Record Linkage Study (an epidemiological database) was used to identify patients who died from a cardiovascular cause within 30 days of emergency or urgent surgery under general anaesthesia. Each case was paired with a control patient (matched for age within 10 yr of the patient, operation and consultant). Additional clinical information was sought from the patient's case notes. Cases and controls were compared for cardiovascular risk factors using conditional logistic regression analysis and a prognostic model was generated. Only one significant risk factor was identified in the final model: a history of cardiac failure (odds ratio 14.84; 95% confidence intervals 2.53-87.13; P = 0.003). Associations between a history of cerebrovascular accident or renal impairment and cardiovascular mortality were seen using univariate analysis but not after adjustment for confounding factors.