Personality predicts occupational performance. However, previous cross-sectional meta-analyses overlook effects of occupational characteristics. Accordingly, we conduct second-order meta-analyses of the Big Five traits and performance in nine major occupational groups: Clerical, customer service, healthcare, law enforcement, management, military, professional, sales, and skilled/semiskilled (k = 539 studies, N = 89,639). For each occupational group, we integrate data from the Occupational Information Network (O*NET): Expert ratings of 1) traits’ relevance to its occupational requirements and 2) its occupational complexity level. We report three major findings. First, trait relations differ considerably across major occupational groups. Second, expert ratings of traits’ relevance largely converge with empirical relations; the top-two most highly rated traits mostly match the top-two most highly predictive traits. Third, complexity moderates performance relations. When occupational groups are ranked by complexity, multiple correlations generally follow an inverse-U shaped pattern, which suggests moderate complexity levels may be the “goldilocks range” for personality prediction.