Responsible dog ownership has been identified as a point of intervention to promote physical activity, based upon an expectation of dog walking in public space. Nevertheless, quantitative research has found variability among owners in their dog walking. In this study, we explore the implications for health promotion of such variability. We do so by drawing on the concepts of habitus and social capital to analyse qualitative interviews. Participants were recruited from a social network in a cosmopolitan city with a policy framework intended to ensure equitable access to public space for dog walkers. The analysis confirms dog ownership can promote both physical activity and social capital, to the extent of mutual reinforcement. Yet we identified patterns of care in which dogs could influence people's emotional well-being without promoting physical activity. In particular, some owners were not capable of extensive dog walking but still benefited emotionally from dog ownership and from interpersonal interactions facilitated by dog ownership. Some participants' dogs, however, could not be walked in public without risking public safety and social sanctions. Responsible dog ownership can therefore also entail not exercising dogs. Contra to the emerging ideal in health promotion, a 'dog-shaped hole' in someone's life does not always take the form of a walking companion.