The distribution of infectious salmon anaemia (ISA) was examined among 80 cages from three Atlantic salmon grow-out farms in Maine, USA that were stocked with smolts from a single hatchery. Cage-level disease was broadly defined as one or more moribund fish testing positive for infectious salmon anaemia virus (ISAV) by RT-PCR and a second confirmatory test (IFAT, culture or genotype sequence). Spatio-temporal and cage-level risks were explored using logistic regression and survival analysis. Non-spatial risk factors associated with ISA, or shortened survival time to disease, included increased predation, trucking company choice for smolt transfers, a finely-sedimented benthic substrate, and smaller average size of smolts at stocking. Univariable analysis identified the time-dependent spatial factor 'adjacency to newly infected cages' to be predictive of new infection in neighbouring cages 11-12 weeks later. However, none of the spatial factors, or their lags retained relevance in multiple-variable models. The results suggest a diffuse distribution of virus exposure throughout infected sites, with host-susceptibility factors probably influencing disease manifestation in individual cages. The narrow focus of the current study may limit application of the findings to other sites and year-classes. However, these data support the relevance of husbandry efforts to optimize fish health in regions affected by ISAV.