AbstractSeed banks represent a reservoir of propagules important for understanding plant population dynamics. Seed viability in soil depends on soil abiotic conditions, seed species, and soil biota. Compared to the vast amount of data on plant growth effects, next to nothing is known about how arbuscular mycorrhizal fungi (AMF) could influence viability of seeds in the soil seed bank. To test whether AMF could influence seed bank viability, we conducted three two‐factorial experiments using seeds of three herbaceous plant species (Taraxacum officinale, Dactylis glomerata, and Centaurea nigra) under mesocosm (experiments 1 and 2) and field conditions (experiment 3) and modifying the factor AMF presence (yes and no). To allow only hyphae to grow in and to prevent root penetration, paired root exclusion compartments (RECs) were used in experiments 2 and 3, which were either rotated (interrupted mycelium connection) or kept static (allows mycorrhizal connection). After harvesting, seed viability, soil water content, soil phosphorus availability, soil pH, and hyphal length in RECs were measured. In experiment 1, we used inoculation or not with the AMF Rhizophagus irregularis to establish the mycorrhizal treatment levels. A significant negative effect of mycorrhizal hyphae on viability of seeds was observed in experiments 1 and 3, and a similar trend in experiment 2. All three experiments showed that water content, soil pH, and AMF extraradical hyphal lengths were increased in the presence of AMF, but available P was decreased significantly. Viability of seeds in the soil seed bank correlated negatively with water content, soil pH, and AMF extraradical hyphal lengths and positively with soil P availability. Our results suggest that AMF can have a negative impact on soil seed viability, which is in contrast to the often‐documented positive effects on plant growth. Such effects must now be included in our conceptual models of the AM symbiosis.