Objective Physical activity (PA) has consistent associations with mental well-being, but studies have focused primarily on leisure-time activity, and there has been little attention to the roles of other activity domains (household, occupational and transport). We examined the dose–response relationships of PA dimensions (frequency, amount and volume) with mental well-being for all four PA domains. We also assessed the interaction effects of gender, age, body weight status, and PA domains. Method In 2003–2004, two surveys collected data on PA, socio-demographics, height and weight, perceived neighborhood attributes, barriers to PA, and physical and mental well-being from 2194 Australian adults. Generalized linear models with restricted cubic splines identified the dose–response relationships of PA domains with mental well-being; the interactive effects of PA domains, age, gender and weight status; and the confounding effects of poor mental or physical health as barriers to PA. Results Leisure-time PA was independently linearly related to mental well-being in most demographic groups. Stronger effects were observed for vigorous-intensity leisure-time PA. Poor health as a barrier to PA explained only a small portion of the relationships of PA with mental well-being. The magnitude and direction of the effects of household, occupational and transport PA depended on age, gender, weight status and/or participation in other PA domains. Conclusions Individual physical capacities and characteristics, and level of discretionary choice are likely determinants of the effects of PA on mental well-being. Strategies aimed at increasing PA for mental health benefits need to take these diverse and sometimes counterintuitive effects into account.