Background: Several studies have found that individuals with a partner were less physically active than those without. To better understand the reasons for this association, we examine whether the existence of a relationship or the current presence of a partner influences physical activity (PA). Methods: We use data from the most recent German Time Use Survey 2012/13 to examine leisure-time PA. All leisure-time activities reported in the time diaries are classified according to their metabolic equivalent of task (MET) to calculate the mean MET scores. First, we use ordinary least square regressions and logistic regressions to examine the effects of living together with a partner or not on mean daily MET scores and on exercise. Second, we apply person-day fixed effects models to estimate the impact of current partner presence on current PA levels. Results: Having a partner is negatively associated with exercising but is not correlated with mean leisure-time MET scores for both women and men. For those with a partner, current MET levels are substantially lower when the partner is present than when the partner is absent. When partners spend leisure-time activities apart, their MET levels are higher than those of individuals without a partner. Conclusions: The results suggest that it is not the mere existence of a romantic relationship but the current copresence with a partner that affects PA behavior. Therefore, interventions to increase PA may be promising if they can encourage couples to be active together.