Despite accumulating evidence pointing to a genetic basis for tardive dyskinesia, results to date have been inconsistent owing to limited statistical power and limitations in molecular genetic methodology. A Medline, EMBASE and PsychINFO search for literature published between 1976 and June 2007 was performed, yielding 20 studies from which data were extracted for calculation of pooled estimates using meta-analytic techniques. Evidence from pooled data for genetic association with tardive dyskinesia (TD) showed (1) in COMT val158met , using Val-Val homozygotes as reference category, a protective effect for Val-Met heterozygotes (OR = 0.63, 95% CI: 0.46-0.86, P = 0.004) and Met carriers (OR = 0.66, 95% CI: 0.49-0.88, P = 0.005); (2) in Taq1A in DRD2, using the A1 variant as reference category, a risk-increasing effect for the A2 variant (OR = 1.30, 95% CI: 1.03-1.65, P = 0.026), and A2-A2 homozygotes using A1-A1 as reference category (OR = 1.80, 95% CI: 1.03-3.15, P = 0.037); (3) in MnSOD Ala9Val, using Ala-Ala homozygotes as reference category, a protective effect for Ala-Val (OR = 0.37, 95% CI: 0.17-0.79, P = 0.009) and for Val carriers (OR = 0.49, 95% CI: 0.24-1.00, P = 0.047). These analyses suggest multiple genetic influences on TD, indicative of pharmacogenetic interactions. Although associations are small, the effects underlying them may be subject to interactions with other loci that, when identified, may have acceptable predictive power. Future genetic research will take advantage of new genomic knowledge.