Fungal strains beneficial for sustainable agriculture, such as those exploited as biocontrol agents (BCAs), cannot be registered and commercialized without providing precise molecular identification data. Existing inaccuracies in DNA databases, such as NCBI GenBank, can cause serious difficulties in this process: false GenBank entries can divert risk assessment and registration processes, and can lead to spurious results concerning the environmental fate of the BCAs and their non-target effects. False GenBank data thus have a negative impact on organic and integrated agricultural programs. This is exemplified by nrDNA ITS sequences wrongly deposited in GenBank for the mycoparasitic genus Ampelomyces, containing well-known commercial BCAs of powdery mildews. Misleading data were published on their presence in the human body, seawater, and other inappropriate environments, where Ampelomyces strains do not colonise. These data were all based on mis-identifications of some common fungal strains with Ampelomyces spp., due to routine methods using erroneous DNA sequences from GenBank. We give examples and propose a new approach for molecular identification processes applied during the assessment of fungal BCAs, and other beneficial organisms, to contribute to their reliable use in sustainable agriculture.