Non-invasive methods of monitoring wild populations (such as genotyping faeces or hair) are now widely used and advocated. The potential advantages of such methods over traditional direct monitoring (such as live capture) are that accuracy improves because sampling of non-trappable individuals may be possible, species in difficult and remote terrain can be surveyed more efficiently, and disturbance to animals is minimal. Few studies have assessed the effects of interactions between species on remote sampling success. We test the use of non-invasive monitoring for the cryptic, forest-dwelling, solitary and endangered bridled nailtail wallaby (Onychogalea fraenata) that is sympatric with the ecologically similar and more common black-striped wallaby (Macropus dorsalis). Six types of hair traps were tested for 3668 trap days, and hairs were caught with about a 10% success rate. Camera traps showed that baited hair traps targeted both wallaby species. We microscopically identified hair as bridled nailtail wallaby or black-striped wallaby. We compared these hairs and their genotypes (using seven microsatellite loci) with known bridled nailtail wallaby hairs and genotypes derived from animal trapping. Trapped bridled nailtail wallaby hairs had characteristics that could be mistaken for black-stripe wallaby hairs; characteristics were not diagnostic. Genetic assignment tests consistently differentiated the known bridled nailtail wallaby samples from identified black-striped wallaby samples, however genetic overlap between most of the microsatellite markers means that they are not suitable for species identification of single samples, with the possible exception of the microsatellite locus B151. With similar trapping effort and within the same area, live-capture mark-recapture techniques estimated 40–60 individuals and non-invasive methods only detected 14 genotypes. A species-specific genetic marker would allow more efficient targeting of bridled nailtail wallaby samples and increase capture rates.