The quantification of the biological diversity in environmental samples using high-throughput DNA sequencing is hindered by the PCR bias caused by variable primer–template mismatches of the individual species. In some dietary studies, there is the added problem that samples are enriched with predator DNA, so often a predator-specific blocking oligonucleotide is used to alleviate the problem. However, specific blocking oligonucleotides could coblock nontarget species to some degree. Here, we accurately estimate the extent of the PCR biases induced by universal and blocking primers on a mock community prepared with DNA of twelve species of terrestrial arthropods. We also compare universal and blocking primer biases with those induced by variable annealing temperature and number of PCR cycles. The results show that reads of all species were recovered after PCR enrichment at our control conditions (no blocking oligonucleotide, 45 °C annealing temperature and 40 cycles) and high-throughput sequencing. They also show that the four factors considered biased the final proportions of the species to some degree. Among these factors, the number of primer–template mismatches of each species had a disproportionate effect (up to five orders of magnitude) on the amplification efficiency. In particular, the number of primer–template mismatches explained most of the variation (~3/4) in the amplification efficiency of the species. The effect of blocking oligonucleotide concentration on nontarget species relative abundance was also significant, but less important (below one order of magnitude). Considering the results reported here, the quantitative potential of the technique is limited, and only qualitative results (the species list) are reliable, at least when targeting the barcoding COI region.