The rationale of this case-control study is to ascertain whether glutamate carboxypeptidase II (GCPII) variants serve as determinants of hyperhomocysteinemia and contribute to the etiology of stroke. Hyperhomocysteinemia was observed in stroke cases compared to controls (14.09 +/- 7.62 micromol/L vs. 8.71 +/- 4.35, P < 0.0001). GCPII sequencing revealed two known variants (R190W and H475Y) and six novel variants (V108A, P160S, Y176H, G206R, G245S and D520E). Among the haplotypes of GCPII, all wild-haplotype H0 showed independent association with stroke risk (OR: 9.89, 95% CI: 4.13-23.68), while H2 representing P160S variant showed reduced risk (OR: 0.17, 95% CI: 0.06-0.50). When compared to subjects with H2 haplotype, H0 haplotype showed elevated homocysteine levels (18.26 +/- 4.31 micromol/L vs. 13.66 +/- 3.72 micromol/L, P = 0.002) and reduced plasma folate levels (7.09 +/- 1.19 ng/ml vs. 8.21 +/- 1.14 ng/ml, P = 0.007). Using GCPII genetic variants, dietary folate and gender as predictor variables and homocysteine as outcome variable, a multiple linear regression model was developed. This model explained 36% variability in plasma homocysteine levels. To conclude, GCPII haplotypes influenced susceptibility to stroke by influencing homocysteine levels.