Most previous research has focused on the relationships between specific personality traits and specific facets of mental health. However, in reality most of the Big Five are associated at non‐trivial levels with mental health. To account for this broad correlation, we proposed the ‘barometer hypothesis’, positing that behind both ratings of mental health and personality lies a barometer that indicates one's general feelings of positivity or negativity. To the extent that both the general factors of personality and mental health reflect this same barometer, we would expect them to be correlated. We tested alternative models using data from a large longitudinal panel study that includes two cohorts of participants who were assessed every two years, resulting in parallel 4‐year longitudinal studies. Similar results were obtained across both studies. Supporting the ‘barometer hypothesis’, findings revealed that the optimal model included general latent factors for both personality traits and mental health. Compared to the broad raw pairwise correlations, the bi‐factor latent change models revealed that the relation among levels and changes in the specific factors were substantially reduced when controlling for the general factors. Still, some relations remained relatively unaffected by the inclusion of the general factor. We discuss implications of these findings. Copyright © 2016 European Association of Personality Psychology