There is often a perceived conflict between allowing recreational access to the countryside and wildlife conservation. Although many studies have investigated potential impacts on birds, few have assessed the potential impacts of recreational disturbance at the population scale. We studied the impact of disturbance on a Woodlark Lullula arborea population on 16 heathland sites in southern England (Mallord et al. 2006). These sites all had historical records of breeding Woodlarks, and together encompassed a range of visitor access levels. A logistic regression model of patch use was used to quantify habitat suitable for Woodlarks. Woodlarks favoured patches with substantial proportions of bare ground and short vegetation. Across sites, Woodlark density (per hectare of suitable habitat) was lower in sites with higher levels of disturbance. Within sites with recreational access, the probability of suitable habitat being colonized was lower in those areas with greater disturbance; this was reduced to below 50% at around eight disturbance events per hour. There was no effect of disturbance on daily nest survival rates. Birds on sites with higher levels of disturbance fledged more chicks (per pair) owing to a strong density-dependent increase in reproductive output. A model is presented (Mallord et al. 2007) that predicts the consequences for the Woodlark population of a range of access scenarios. The model suggested that recreational disturbance is already having a major effect on the Woodlark population. Additionally, it predicted that any further impact on the population depended on both the numbers of people and their spatial distribution. Under current access arrangements, a doubling of visitor numbers had little effect, whereas the same number of people distributed evenly across all sites led to a major negative impact on the population. Many previous studies have stressed the negative effects of recreational disturbance on bird behaviour, distribution and breeding success. However, from a conservation viewpoint, the impact at the population level is of paramount importance. Modelling the consequences of alternative access scenarios will help policymakers in the development of appropriate mitigation measures.