The serotonin transporter (5-HTT) is a candidate gene for bipolar disorder (BPD). It has been investigated for association with the illness in a series of studies, but overall results have been inconsistent and its role in the disorder remains controversial. Systematic reviews using meta-analytical techniques are a useful method for objectively and reproducibly assessing individual studies and generating combined results. We performed two meta-analyses of published studies--both population-based and family-based studies--investigating the association between BPD and the 5-HTT gene-linked polymorphic region (5-HTTLPR) and the intron 2 variable numbers of tandem repeats (VNTR) polymorphisms. The literature was searched using Medline and Embase to identify studies for inclusion. We statistically joined population-based and family-based studies into a single meta-analysis. For both polymorphisms, our review revealed significant pooled odds ratios (ORs): 1.12 (95% CI 1.03-1.21) for the 5-HTTLPR and 1.12 (95% CI 1.02-1.22) for the intron 2 VNTR. Meta-regression showed that neither the study type (population-based vs family-based; P=0.41 for the 5-HTTLPR and P=0.91 for the intron 2 VNTR) nor the sample ethnicity (Caucasian vs non-Caucasian; P=0.35 for the 5-HTTLPR and P=0.66 for the intron 2 VNTR) significantly contributed to the heterogeneity of the meta-analyses. The observed ORs could be regarded simply as a very small but detectable effect of the 5-HTT, which has an additive effect when combined with other susceptibility loci. Alternative hypotheses on this finding were also discussed: a stronger effect of the haplotypes involving the two polymorphisms or other SNP markers; a more direct effect of these polymorphisms on specific phenotypes of BPD; and the presence of gene-environment interaction as a mediator of the genetic effects of 5-HTT.