Abstract.Obtaining useful information about elusive or endangered species can be logistically difficult, particularly if relying entirely on field signs such as hair, feathers or faeces. However, recent developments in molecular technology add substantially to the utility of such 'non-invasive' samples, which provide a source of DNA that can be used to identify not only species but also individuals and their gender. This provides great potential to improve the accuracy of abundance estimates and determine behavioural parameters, such as home-range size, individual habitat and dietary preferences, and even some forms of social interaction. Non-invasive samples can also be a useful alternative to blood or tissue samples (the collection of which traditionally has required trapping of animals) as genetic material for applications such as relatedness, population genetic and phylogenetic analyses. Despite the huge potential of non-invasive genetic sampling, the current technology does have limitations. The low quantity and quality of DNA often obtained from such sources results in an increased risk of genotyping errors, which may lead to incorrect inferences, particularly false identification of individuals. Appropriate precautions and pilot studies are required to minimise these risks, and in some cases it may be wise to employ traditional methods when they are adequate.