BACKGROUND:An inverse relationship between hospital volume and death following pancreatico duodenectomy (PD) has been reported from several countries. The aim of this study was to assess the volume-outcome effect of PD in Italy. METHODS:The study group comprised 1576 patients who underwent PD in 2003. Hospitals were allocated to four volume groups: low volume, five PDs or fewer; medium volume, six to 13 PDs; high volume, 14 to 51 PDs; and very high volume, two hospitals that performed 89 and 104 PDs. RESULTS:Some 221 hospitals performed at least one PD in 2003; hospital volume was low in 74.7 per cent, medium in 17.6 per cent, high in 6.8 per cent and very high in 0.9 per cent. The overall mortality rate was 8.1 per cent. Increasing hospital volume was associated with a significantly reduced mortality rate: 12.4 per cent (adjusted odds ratio (OR) 1.000) for low-volume, 7.8 per cent (OR 0.611) for medium-volume, 5.9 per cent (OR 0.466) for high-volume and 2.6 per cent (OR 0.208) for very high-volume hospitals. Length of postoperative stay was reduced in very high-volume hospitals (P < 0.001). CONCLUSION:The outcome of PD in Italy is dependent on hospital volume and a policy of centralization may therefore be appropriate.