Causality assessment is an algorithm proposed by WHO to identify a causal relationship between vaccines \nand adverse events following immunization (AEFIs), mostly for serious adverse events. It can be \nconsidered consistent, inconsistent, indeterminate or unclassifiable. \nThis study describes AEFIs reported in Puglia from 2013 to 2016 and analyzes the differences between \nthe causality assessments performed on AEFI case-report information and the causality assessments \nperformed after the examination of clinical documentation. \n292 AEFI were reported: 191 (65.4%) non serious, 59 (20.2%) serious and 42 (14.4%) undefined. \nCausality assessment performed on the AEFI case-report information classified 59.2% (n=29/49) of \nserious AEFIs as consistent while assessment performed after clinical review only classified 30.6% \n(n=15/49) of serious AEFI as consistent (X2=65.0; p=0,000). In the first approach, inconsistent serious \nAEFIs were 18.6% (n=11/49) and then became 45.8% (n=27/49) after examination of clinical \ndocumentation. Indeterminate serious AEFIs were 6.8% (n=4) at first, and then 3.4% (n=2). \nUnclassifiables did not change.