Predation by generalist predators is difficult to study in the field because of the complex effects of positive and negative interactions within and between predator species and guilds. Predation can be monitored by molecular means, through identification of prey DNA within predators. However, polymerase chain reaction (PCR) amplification of prey DNA from predators cannot discriminate between primary and secondary predation (hyperpredation), in which one predator feeds on another that has recently eaten the target prey. Here we quantify, for the first time, the potential error caused by detection of prey DNA following secondary predation, using an aphid-spider-carabid model. First, the aphid Sitobion avenae was fed to the spider Tenuiphantes tenuis and the carabid Pterostichus melanarius, and the postconsumption detection periods, for prey DNA within predators, were calculated. Aphids were then fed to spiders and the spiders to carabids. Aphid DNA was detected in the predators using primers that amplified 245- and 110-bp fragments of the mitochondrial cytochrome oxidase I gene. Fragment size and predator sex had no significant effect on detection periods. Secondary predation could be detected for up to 8 h, when carabids fed on spiders immediately after the latter had consumed aphids. Beetles tested positive up to 4 h after eating spiders that had digested their aphid prey for 4 h. Clearly, the extreme sensitivity of PCR makes detection of secondary predation more likely, and the only reliable answer in future may be to use PCR to identify, in parallel, instances of intraguild predation.