In 2022 a large-scale test of a commercial deep-sea mining machine was undertaken on the abyssal plain of the eastern Pacific Ocean at a depth of 4,280 m, recovering over 3,000 t of polymetallic nodules. Here, using a quantitative species-level sediment-dwelling macrofaunal dataset, we investigated spatio-temporal variation in faunal abundance and biodiversity for 2 years before and 2 months after test mining, allowing separation of direct mining impacts from natural background variation. Macrofaunal density decreased by 37% directly within the mining tracks, alongside a 32% reduction in species richness, and significantly increased community multivariate dispersion. While species richness and diversity indices within the tracks were reduced compared with controls, diversity was not impacted when measured by sample-size independent measures of accumulation. We found no evidence for change in faunal abundance in an area affected by sediment plumes from the test mining; however, species dominance relationships were altered in these communities, reducing overall biodiversity. These results provide critical data on the effective design of abyssal baseline and impact surveys and highlight the value of integrated species-level taxonomic work in assessing the risks of biodiversity loss. Deep-sea mining within the CCZ is at a critical juncture, as the industry looks to move beyond the exploration phase and into commercial exploitation. Consequently, direct assessment of mining impacts on faunal abundance and biodiversity at the seafloor is needed. Immediately following impact, macrofaunal densities decreased significantly by 37% within the path of the nodule collector, while densities within the control sites either increased or remained unchanged, aligning with patterns from previous smaller disturbance experiments showing 38–63% decreases in macrofaunal density. Parallel with the observed reduction in faunal abundance, a significant decrease in species richness and diversity within the mining tracks was observed, with the significantly lower abundance within the tracks likely driving reduced diversity per sample unit.