The molecular evolution of feline calicivirus (FCV) was studied in cell culture and in persistently infected cats. Sequence analysis of the 5' hypervariable region of the FCV capsid (5'HVR; located at the 5'end of variable region E), a region known to contain linear neutralising B cell epitopes, showed FCV existed as a quasispecies which evolved at the nucleotide and amino acid level during persistent infection. Quasispecies heterogeneity tended to decrease during the course of persistence. Sequential isolates from a cat showed marked antigenic variation during the course of persistent infection. Sequential passage of FCV in cell culture was also associated with sequence evolution of the 5'HVR. However, these isolates showed no change in antigenicity suggesting that individual substitutions observed in viruses from cats, but not in viruses from cell culture, may be responsible for changes in antigenicity. Alternatively, the observed antigenic changes may be associated with mutations elsewhere in the genome. In order to identify regions of 'the FCV capsid protein containing linear B-cell epitopes, two approachesw ere used.F irstly, an expressionl ibrary containingr andom, short (100- 300bp) fragments of an FCV capsid gene was constructed. This library was screened using polyclonal antisera from a cat that had been challenged experimentally with FCV to identify immunoreactive clones containing B-cell epitopes. Initial screening identified five clones that reacted positively to feline antisera in immunoblots. FCV derived sequencefr om thesec lones all mappedt o the 5'HVR suggestingt his region containst he immunodominant linear epitopes of the capsid. The second approach used to identify B-cell epitopes was to map more accurately the epitope of a neutralising monoclonal antibody (IG9) which had already been shown to lie in a 37 amino acid region of the 5'HVR (Milton et al. (1992), Journal of General Virology 73,2435-2439). Replication of plaque purified IG9-sensitive parent virus in sub-neutralising concentrations of IG9 led to the generation of a neutralisation resistant escape mutant. Sequence analysis of this mutant and the parent virus revealed a single non-synonymous nucleotide substitution within the 5'HVR suggesting this residue is critical to the correct formation of the IG9 epitope. A method to type FCVs based on sequence analysis of the 5'HVR was established. Most isolates appeared relatively homogenous. However, some isolates, both from vaccines and vaccine failures, appeared to contain more than one FCV. Comparison of 5'HVR sequencesfr om different isolatess howed that most isolates were either 0-5.3% different (related isolates) or 20.7-42.7% different (unrelated isolates). The majority of the relatedi solatess hareda n epidemiologicall ink, implying they representedis olatest hat originated from a common source. Comparisons of sequences obtained from vaccine failures and vaccine virus fell into two similar categories; those with closely related sequences(0 .0-5.3%) implying a role for the vaccinei n diseasea nd those with divergent sequences(2 1.3-38.7%)i mplying field virus causedt he disease. These results were compared with those obtained by using a serotyping method based on virus neutralisation (VN) which exploits differences in antigenicity between most FCVs gene (Dawson et al. (1993), Veterinary Record 132,346-350). VN and sequence analysis gave the same typing result in 65-73% of individual cases. Based on these results and the difficulty of interpreting VNs, we suggest that molecular based sequence analysis may be more suitable to the epidemiological investigation of FCV related disease particularly in the case of vaccine reactions.