We apply qPCR molecular techniques to detect in situ rates of consumption of sea urchins (Centrostephanus rodgersii and Heliocidaris erythrogramma) by rock lobsters (Jasus edwardsii). A non-lethal method was used to source faecal samples from trap-caught lobsters over 2 years within two no-take research reserves. There was high variability in the proportion of lobsters with faeces positive for sea urchin DNA across years and seasons dependent on lobster size. Independent estimates of lobster predation rate on sea urchins (determined from observed declines in urchin abundances in the reserves relative to control sites) suggest that rates of molecular prey detection generally overestimated predation rates. Also, small lobsters known to be incapable of directly predating emergent sea urchins showed relatively high rates of positive tests. These results indicate that some lobsters ingest non-predatory sources of sea urchin DNA, which may include (i) ingestion of C. rodgersii DNA from the benthos (urchin DNA is detectable in sediments and some lobsters yield urchin DNA in faeces when fed urchin faeces or sediment); (ii) scavenging; and/or predation by rock lobsters on small pre-emergent urchins that live cryptically within the reef matrix (although this possibility could not be assessed). While the DNA-based approach and direct monitoring of urchin populations both indicate high predation rates of large lobsters on emergent urchins, the study shows that in some cases absolute predation rates and inferences of predator-prey interactions cannot be reliably estimated from molecular signals obtained from the faeces of benthic predators. At a broad semi-quantitative level, the approach is useful to identify relative magnitudes of predation and temporal and spatial variability in predation.