AbstractCharacterisation of new viruses is often hindered by difficulties in amplifying them in cell culture, limited antigenic/serological cross‐reactivity or the lack of nucleic acid hybridisation to known viral sequences. Numerous molecular methods have been used to genetically characterise new viruses without prior in vitro replication or the use of virus‐specific reagents. In the recent metagenomic studies viral particles from uncultured environmental and clinical samples have been purified and their nucleic acids randomly amplified prior to subcloning and sequencing. Already known and novel viruses were then identified by comparing their translated sequence to those of viral proteins in public sequence databases. Metagenomic approaches to viral characterisation have been applied to seawater, near shore sediments, faeces, serum, plasma and respiratory secretions and have broadened the range of known viral diversity. Selection of samples with high viral loads, purification of viral particles, removal of cellular nucleic acids, efficient sequence‐independent amplification of viral RNA and DNA, recognisable sequence similarities to known viral sequences and deep sampling of the nucleic acid populations through large scale sequencing can all improve the yield of new viruses. This review lists some of the animal viruses recently identified using sequence‐independent methods, current laboratory and bioinformatics methods, together with their limitations and potential improvements. Viral metagenomic approaches provide novel opportunities to generate an unbiased characterisation of the viral populations in various organisms and environments. Copyright © 2007 John Wiley & Sons, Ltd.