ABSTRACT: Little is known about the dose-response relationship between recreational arts engagement (for enjoyment, entertainment or as a hobby, rather than therapy) and mental well-being in the general population. The quantification of this relationship is of value to: (1) health professionals, clinicians and researchers interested in utilising the arts as a method for improving mental health; (2) to health promoters and policy makers in the development of population based health messages, policy and practice; and (3) to members of the general public in maintaining or improving their own well-being. As guided by theories of social epidemiology and the biopsychosocial model of health, the first aim of this study was to determine if there was a relationship between arts engagement (hours per year) and mental well-being in the general population. If an association was demonstrated, the second aim was to quantify this relationship.A random sample of 702 Western Australian adults aged 18+ years (response rate = 71%) were invited to take part in a telephone survey. The survey took 15 min to complete and included questions about arts engagement, mental well-being, demographics and potential confounders/effect modifiers. The dependent variable was subjective mental well-being (Warwick-Edinburgh Mental Well-being Scale, WEMWBS). The independent variable was hours engaged in the arts in the last 12 months.Respondent engagement in the arts ranged from zero to 1572 hours/year (mean = 100.8 hours/year, SD = 206.0). The prevalence of engagement was 83 %. The average WEMWBS score was 53 (SD = 7.4). After adjustment for demographics (i.e. sex, age group, location, income, education, marital status, children), general health, sports engagement, religious activities and holidays, respondents with high levels of arts engagement (100 or more hours/year, WEMWBS score = 55) had significantly better mental well-being than those with none (0 hours/year, WEMWBS score = 53), low (0.1-22.9 hours/year, WEMWBS score = 52) and medium (23-99.9 hours/year, WEMWBS score = 53) levels of engagement (p = 0.003). Respondents with none, low and medium arts engagement had similar WEMWBS scores (p = 0.358). The relationship between arts engagement and WEMWBS was nonlinear with evidence of a minimum threshold at 100 or more hours/year (p = 0.0006).Evidence of an arts-mental health relationship was found in this study. Those who engaged in 100 or more hours/year of arts engagement (i.e. two or more hours/week) reported significantly better mental well-being than other levels of engagement. The suitability of the arts as a population based strategy to influence the mental well-being of the general population should be investigated further.