Disease dynamics hinge on parasite transmission among hosts. However, canonical models for transmission often fit data poorly, limiting predictive ability. One solution involves building mechanistic yet general links between host behaviour and disease spread. To illustrate, we focus on the exposure component of transmission for hosts that consume their parasites, combining experiments, models and field data. Models of transmission that incorporate parasite consumption and foraging interference among hosts vastly outperformed alternatives when fit to experimental data using a zooplankton host (Daphnia dentifera) that consumes spores of a fungus (Metschnikowia bicuspidata). Once plugged into a fully dynamic model, both mechanisms inhibited epidemics overall. Foraging interference further depressed parasite invasion and prevalence at high host density, creating unimodal (hump-shaped) relationships between host density and these indices. These novel results qualitatively matched a unimodal density–prevalence relationship in natural epidemics. Ultimately, a mechanistic approach to transmission can reveal new insights into disease outbreaks.