We attempted to replicate the findings of two recent meta-analyses that personality inventory moderates the association between the serotonin transporter gene and anxiety-related traits. A total of 24 studies contributed to the meta-analysis, of which three reported genotype frequencies that deviated from Hardy-Weinberg (HW) equilibrium. We found some support for the view that results depend on the type of questionnaire used, although in a direction opposite to that previously reported. Contrasts between the S/S and L/L groups were significant for TCI/TPQ harm avoidance studies (P ¼ 0.0024) but not NEO neuroticism (P ¼ 0.9757). When studies not in HW equilibrium were excluded the TCI/TPQ result for the S/S genotype still exceeded our 5% threshold, although with reduced significance (P ¼ 0.0082), and the NEO result remained nonsignificant (P ¼ 0.9109). While we cannot rule out an association between the 5HTT gene and anxiety-related traits, particularly for TCI/TPQ harm avoidance, our findings do indicate that the effect, if present, is small. Our results emphasise the importance of complete ascertainment of studies and the identification of relevant sources of heterogeneity. Molecular Psychiatry ( It is now accepted that a substantial proportion of variability in human personality is attributable to genetic variation. 1 As theories of personality propose that individual differences in specific traits are associated with individual differences in specific neurotransmitters, a number of candidate genes with known functional effects on neurotransmitter pathways have been investigated. 2 One pathway that has received considerable attention is the serotonin system, which is known to play an important role in emotional behaviour. 3 However, attempts to demonstrate an association between variation in genes involved in serotonin function and neuroticism or harm avoidance, a personality trait believed to reflect inherited differences in emotional response, have yielded inconsistent results.Even three independent meta-analyses of the large number of relevant publications do not come to a consensus: while Munafò et al 4 found little evidence for an association, both Schinka et al 5 and Sen et al 6 reported that 5HTT-LPR was significantly associated with neuroticism as measured using the class of personality questionnaires derived from Costa and McCrae's 7 five-factor model of personality (NEO-PI, NEO-PI-R, NEO-FFI), but was not associated with harm avoidance as measured using the class of questionnaires derived from Cloninger's 8 tridimensional theory of personality (TCI, TPQ).The findings of the Schinka and Sen meta-analyses are an important methodological and conceptual challenge to personality genetics. NEO neuroticism demonstrates substantial psychometric form equivalence with TCI/TPQ harm avoidance 9 and, assuming that questionnaire equivalence reflects the measurement of the same underlying biological substrate, each assessment of neuroticism should be associated with the same genetic variants. Failure to provide com...