Herbivores have incomplete knowledge of their local environment which creates a two stage grazing process within heterogeneous systems i.e. patch selection based on visual cues (e.g. sward height) and patch rejection based on more localized cues (e.g. olfactory cues associated with faeces). The relative strength of these cues will determine the grazing decisions (i.e. patch choice) of herbivores which will determine their nutrient intake rate, subsequent sward structure and thus the efficiency of use of the forage resource. In addition, the trade-off between faecal avoidance and the desire to maximise intake will control the risk of exposure to faecally transmitted disease. Building on previous work we further develop an agent-based modelling framework used to describe grazing and avoidance behaviour in a spatially explicit context. The framework is based on discrete state-space Markov processes which provide general and flexible methods both to describe and infer the behaviour of a broad range of systems. We explore potential disease transmission risks from both grazing contacts and investigative behaviour, and discuss their relative timing and importance. In addition we explore the impact of different distributions of contamination on disease exposure and explore the sensitivity of such risks to assumptions about the distance over which the animals search.