Herbivores grazing in fenced areas risk high parasite-loads when forced to graze near their own dung. We investigated during a 4-month period (July, August, September and October) the influence of parasite-infected dung on grazing behaviour, patch selection and parasite-load of horses. A natural group of eleven (one stallion, 5 mares and 5 foals) free-ranging, naturally parasite-infected Dartmoor ponies of varying age, sex and parasite-load were observed in an area of 4.3 hectares of grassland. Behaviour (including grazing, etc) and distance to nearby dung (7 distance categories: 0-1m, 1-2m, 2-3m, 3-4m, 4-5m, 5-10m, 10+m) were recorded. Faecal Egg Counts (FEC) of 4 dung samples (2 in August and 2 in October) of all individuals were collected and analysed. During October dung densities were experimentally manipulated; in areas selected for low dung densities, white spots were painted. After 4 days fresh dung was placed on the spots in half of these areas. Behaviour and distance from spots (4 days) and dung (4 days) were registered using scan sampling (5-min interval). Data were analysed with Chi-square and GLM tests. The horses grazed significantly less than expected (Chi-square=123.48, df=1, P<0.001) within a short distance (0-1m) from dung, and significantly more than expected (Chi-square=239.06, df=1, P<0.001) four or more metres away from dung. FEC scores for Strongylus spp. (blood worm) showed a positive and significant correlation with individual grazing activity (in %) at 0-1m distance from dung piles (Rs=0.76, P=0.011, N=11). After addition of faeces to low density dung areas, the grazing activity decreased from 45% to 29% (F=13.34, df=1,7, P=0.008). Dartmoor ponies avoid grazing near dung piles and avoid also areas with high dung densities. The avoidance is not complete, and individuals that graze more in areas 0-1m from dung piles have a higher parasite-load and hence may run more health and welfare risks.