Although higher physical activity (PA) levels are associated with better mental health, previous findings about the shape of the dose–response relationship between PA and mental health are inconsistent. Furthermore, this association may differ according to sedentary levels. We investigated the cross-sectional dose-response associations between objectively measured PA and mental health in a representative national sample of adults. We also examined whether sedentary time modified the PA – mental health associations. Based on 2007-2013 Canadian Health Measures Survey data, PA and sedentary time were measured using accelerometry among 8150 participants, aged 20 to 79 years. Generalized additive models with a smooth function were fitted to examine associations between minutes per day of moderate and vigorous PA (MVPA), light PA (LPA), daily steps (combined or not with sedentary time) and self-rated mental health. A significant curvilinear relationship between average daily minutes of MVPA and mental health was observed, with increasing benefits up to 50 minutes/day. For LPA, a more complex shape (monotonic and curvilinear) was found. For daily steps, inverted U-shaped curve suggested increasing benefits until a plateau between 5000 and 16000 steps. The MVPA-LPA combination was significantly associated with mental health but with a complex pattern. The tested PA-sedentary time combinations showed that increasing sedentary time decreased the positive PA-mental health associations. Non-linear dose-response patterns between the PA modalities and self-reported mental health were observed. Optimal doses of daily minutes of MVPA, LPA, MVPA combined with LPA and daily steps are independently associated with better mental health in adults. The results also suggest that PA-mental health associations could be hampered by daily sedentary time.