By analysing the nucleotide sequence data generated from both the E2 (gp55) and the NS5B genes of classical swine fever virus (CSFV), in addition to previously published data from the 5'NCR, we were able to divide 115 CSFV isolates into two major groups, five subgroups and two disparate isolates. Further discrimination was possible by analysis of sequence data from the E2 region. The three sequencing based methods were compared to monoclonal antibody (MAb) typing and to limited restriction enzyme (RE) mapping. Although both MAb and RE methods confirmed the previous classification the resolution was inferior. We estimated an approximate evolution rate for CSFV from an analysis of the virus variation observed in a single geographical area over a 6 year period. Applying this proposed rate to each of our deduced CSFV subgroups enabled us to calculate the approximate dates of divergence for each subgroup.