During a period in which vaccination of splenectomized patients has been recommended, we analysed the patterns of severe post-splenectomy infections (i.e. bacteraemia or meningitis) in a defined population-based cohort. A total of 561 patients undergoing splenectomy were identified during 1984-93 in a Danish county, and the 538 eligible patients were followed for 1731 person-years. After splenectomy, 38 patients contracted a bacteraemia, of which 45% occurred within 30 d (i.e. during the postoperative period). No cases of meningitis were found. Among splenectomized patients the incidence rate of bacteraemia was 2.3 per 100 person-years at risk. Beyond the postoperative period we found an 8-fold increased risk of bacteraemia. Enterobacteria were the predominant cause (45%), and only 1 case due to Streptococcus pneumoniae was recorded. 89 (17%) died during the postoperative period, and the overall mortality rate was 18.4 per 100 person-years at risk. In all, 60% of the patients had been given a pneumococcal vaccination, and a Cox proportional hazard regression model showed that vaccination significantly reduced the risk of bacteraemia of any cause beyond the postoperative period. We conclude that splenectomy increases the risk of severe infections, and that vaccinated patients carry a lower risk of infection than non-vaccinated ones.