Numerous agents of infections in humans and other mammals are found among fungi that are able to survive extreme environmental conditions and to quickly adapt to novel habitats. Nevertheless, the relationship between opportunistic potential and polyextremotolerance was not yet studied systematically in fungi. Here, the link between polyextremotolerance and opportunistic pathogenicity is shown in a kingdom-wide phylogenetic analysis as a statistically significant co-occurrence of extremotolerance (e.g. osmotolerance and psychrotolerance) and opportunism at the level of fungal orders. In addition to extremotolerance, fungal opportunists share another characteristic—an apparent lack of specialised virulence traits. This is illustrated by a comparative genomic analysis of 20 dothideomycetous and eurotiomycetous black fungi. While the genomes of specialised fungal plant pathogens were significantly enriched in known virulence-associated genes that encode secreted proteases, carbohydrate active enzyme families, polyketide synthases, and non-ribosomal peptide synthetases, no such signatures were observed in human opportunists. Together the presented results have several implications. If infection of human hosts is a side effect of fungal stress tolerance and adaptability, the human body is most likely neither the preferred habitat of such species, nor important for their evolutionary success. This defines opportunism as opposed to pathogenicity, where infection is advantageous for the species’ fitness. Since opportunists are generally incapable of the host-to-host transmission, any host-specific adaptations are likely to be lost with the resolution of the infection, explaining the observed lack of specialised virulence traits. In this scenario opportunistic infections should be seen as an evolutionary dead end and unlikely to lead to true pathogenicity.