Previous cross-sectional and intervention studies have suggested that pet owners may enjoy better physical and mental health than non-owners. This paper presents longitudinal evidence from a major national representative longitudinal survey: the German Socio-Economic Panel (SOEP). Because the data are longitudinal, it is possible to assess the impact on health outcomes (measured by number of doctor visits) of longer term pet ownership, and also of gaining and losing a pet. An unexpected finding was that all health benefits appear to accrue to homeowners only. The main result, then, is that homeowners who have owned a pet for five years or more make significantly fewer doctor visits than non pet owners. However, losing a pet appears to impose immediate health costs. The results hold after controlling for other variables associated with use of health services, and also for health status at baseline. They still hold when a proxy for unobserved heterogeneity is included in equations.