A variable region of the gene encoding the major glycoprotein (E2) of Classical Swine Fever Virus (CSFV) was sequenced from 12 Sardinian isolates which had been obtained from three geographically distinct regions of the Island. Phylogenetic analysis of these viruses and others characterized in previous studies [1, 2] indicated that (a) the Sardinian viruses were all members of the common European subgroup 2·3 and were clearly distinct from live vaccines recently used in this area; (b) they could be resolved into four distinct groups in accordance with the region or date of isolation; (c) in at least two regions wild boar/domestic swine contact was implicated in virus spread; (d) the oldest isolate (1983) and some of the recent isolates were possibly introduced from mainland Italy. In addition, this study has wider implications for the interpretation of CSFV variation. We have been able to demonstrate that small variations within this region of the virus genome (possibly less than 2·7% or five nucleotide substitutions) can be used to separate isolates into groups that precisely fit their geographical distribution. This finding is especially important for deducing the epidemiological relationships between multiple outbreaks caused by similar viruses that occur in close proximity.