CONTEXT: Variation at the insulin gene VNTR (variable number tandem repeat) minisatellite has been reported to be associated with polycystic ovary syndrome (PCOS), but findings have been inconsistent and all studies have featured small sample sizes. OBJECTIVE: To gain a robust understanding of the role of the INS-VNTR in PCOS susceptibility. DESIGN: Case-control, family-based association and quantitative trait analyses. SETTING AND PARTICIPANTS: A UK population comprising 255 parent-offspring trios, 185 additional cases, and 1062 control subjects (cases and controls all British/Irish) as well as 1599 women from a northern Finland population-based birth cohort characterized for PCO symptomatology and testosterone levels. VNTR class was inferred from genotyping of the -23HphI variant. INTERVENTION(S): None. MAIN OUTCOME MEASURE(S): INS-VNTR genotype frequencies between subject groups, body mass index, and testosterone levels by genotype. RESULTS: Case-control analyses in both UK and Finnish samples failed to confirm previously reported class III allele associations with PCOS (UK, P = 0.43, Finnish, P = 0.31; Kruskal-Wallis chi2). Transmission analysis in trios showed no excess transmission of either allele (P = 0.62), regardless of parent of origin (maternal: P = 0.73; paternal: P = 0.66). No association between genotype and testosterone levels was seen in any sample (UK PCOS subjects, P = 0.95; Finnish symptomatic cases, P = 0.38; Finnish control women, P = 0.58). CONCLUSIONS: Despite the strong biological candidacy and supportive data from previous studies, we conclude that variation at the INS-VNTR has no major role in the development of PCOS.