The evidence is conflicting as to whether body mass index (BMI) is associated with mental health and, if so, to what extent it varies by sex and age. We studied mental health across the full spectrum of BMI among the general population, and conducted subgroup analyses by sex and age.

We undertook a cross-sectional study of a representative sample of the Scottish adult population. The Scottish Health Survey provided data on mental health, measured by the General Health Questionnaire-12 (GHQ), BMI, demographic and life-style information. Good mental health was defined as a GHQ score <4, and poor mental health as a GHQ score ⩾4. Logistic regression models were applied.

Of the 37 272 participants, 5739 (15.4%) had poor mental health. Overall, overweight participants had better mental health than the normal-weight group [adjusted odds ratio (OR) 0.93, 95% confidence interval (CI) 0.87–0.99, p = 0.049], and individuals who were underweight, class II or class III obese had poorer mental health (class III obese group: adjusted OR 1.26, 95% CI 1.05–1.51, p = 0.013). There were significant interactions of BMI with sex (p = 0.013) and with age (p < 0.001). Being overweight was associated with significantly better mental health in middle-aged men only. In contrast, being underweight at all ages or obese at a young age was associated with significantly poorer mental health in women only.

The adverse associations between adiposity and mental health are specific to women. Underweight women and young women who are obese have poorer mental health. In contrast, middle-aged overweight men have better mental health.

To investigate the associations between obesity and self-rated health and psychological well-being in Spanish women.

Cross-sectional study. Three dependent variables were defined: self-rated health; self-declared diagnosis of psychiatric disorders or use of psychiatric drugs; and General Health Questionnaire-12 (GHQ-12) score. A set of variables (sociodemographic, morbidity and lifestyle) were used to adjust for possible confounding effects.

The National Health Survey was conducted in Spain in 2006.

A total of 15 099 women aged ≥18 years. Participants were classified into groups according to their BMI.

In all, 55·4 % of the women had normal weight, 29·4 % were overweight and 15·2 % were obese. Self-perception of poor health in obese women was 57·8 %, a significantly higher value than in women of normal weight (28·8 %). Prevalence of psychiatric disease was 35·5 % in obese women and 18·9 % in women of normal weight. In multivariate analysis, obese women had 34 % higher odds of declaring poor self-perception of health (OR = 1·34; 95 % CI 1·12, 1·61), 18 % higher odds of self-reporting psychiatric disease (OR = 1·18; 95 % CI 1·02, 1·38) and 26 % higher odds (OR = 1·26; 95 % CI 1·02, 1·55) of having an abnormal outcome (≥3) on the GHQ-12 than women of normal weight.

The present study highlights that obese Spanish women have worse self-rated health and psychological health than those with normal weight. These aspects are relevant because of the growing importance placed on the functionality of patients and their mental health within the obesity epidemic.