We examined habitat use in a population of synurbic watersnakes with equal access to urban and natural habitats to test the hypotheses that species occupy urban environments either by (1) restricting their activities to any remaining natural areas, or (2) capitalizing on, instead of avoiding, artificial features. For three years we radio-tracked 50 northern watersnakes (Nerodia sipedon) living in a 40-ha area along 2 km of a city stream in Pennsylvania (USA). Half the study site is urbanized (municipal park and an active industrial area), and half is a relatively natural conservation area. Habitats selected by snakes in the two areas were significantly different: in the natural half, snakes occupied areas with a wide riparian zone and dense canopy cover; in the urban half, they frequently used artificial substrates and were in close proximity to people. Snakes were relocated 2520 times, yet were found at only 113 sites. Frequently reused sites were mostly artificial, including piles of scrap metal or concrete, holes in a railroad bed adjacent to the stream, and dead evergreen trees secured into the stream bank to combat erosion. Urban and natural areas were approximately equal in area and stream length, and had similar numbers of snake-selected sites (64 urban, 49 natural), but urban sites were used by more snakes. Of sites used by more than five different snakes, 22 of 26 were in the urban area. Snakes were found within 5 m of a tagged conspecific at 38% of urban area relocations compared to 15% of natural area relocations. These data suggest that anthropogenic structures in urban environments provide conditions (concealment, thermal) that offset dangers posed by closer proximity to people.