Habitat use was studied for 28 deer (Odocoileus virginianus) in a suburban area 20 km northeast of the city of Philadelphia. Deer were tracked at high frequency by the use of GPS collars (one GPS fix every 5 min for weeks to months) providing high resolution location data. Habitat disturbance was ranked using several approaches. One metric was based on a human activity index (scale 0–10, with zero corresponding to least impacted and ten to most impacted habitat). A second metric was based on vegetation succession depression [scale 0–5, with zero indicating areas undergoing natural succession and five for areas where succession is almost impossible (paved roads, buildings)]. In addition habitat was classified by the density of occupied buildings. The resulting GIS model depicting habitat disturbance (Disturbance Index, DI) was then plotted against GPS fixes generated by individual deer. Monitored animals selected wooded habitat with shrubby underbrush, neglected land parcels, and fields with annual and bi-annual mowing regimes. They avoided high-density residential areas and very low disturbance areas, but tolerated areas with low density of buildings (fewer than 400 buildings km⁻²). Deer preferences along the suburban disturbance gradient suggested that patches of moderate disturbance with ongoing succession in areas with low building density are most important for the deer. Managing these areas could be instrumental in controlling the deer herd.