Nuclear mitochondrial insertions (NUMTs) are sequences homologous to mtDNA, which are present throughout the human nuclear genome. The possibility that these sequences may be accidentally amplified in reactions directed to mtDNA has been raised and evaluated by different groups and by different means. Despite that, data is still missing on the specificity of PCRs in routine procedures in what concerns contamination with nuclear mtDNA insertions (NUMTs). In this work, we performed PCR sequencing reactions with primers directed either to mitochondrial or to NUMT DNA with different annealing temperatures and in different tissues. We observed that (a) contamination with NUMTs depends on the sample and tissue, and (b) employing routine techniques, there is no risk of co-amplification. Only when mtDNA is almost completely removed from the samples does the number of NUMT copies exceed mitochondrial sequences, i.e., only in samples with virtually no mtDNA, such as those resulting from preferential semen lysis, is there a risk of accidental amplification of NUMTs. We suggest that to evaluate a possible co-amplification of NUMT DNA, it is more relevant to take into account sample processing and original tissue of the samples, and consequently the relative proportions of NUMT and mtDNA, rather than the presence of NUMTs by itself, irrespectively of its proportion.