Abstract.Aims/hypothesis: Vitamin D, a molecule with antiproliferative, antiangiogenic, antioxidant and immunosuppressive effects, could play a role in the pathogenesis of severe diabetic retinopathy. We examined whether Taq I polymorphism of the vitamin D receptor is involved in the development of severe diabetic retinopathy. Methods: 200 unrelated C-peptide-negative French Type I diabetic patients were randomly selected (male:female, 103:97, age 44.4 ± 12.4 years, diabetes duration: 27.7 ± 10.0 years, BMI: 24.3 ± 3.4 kg/m2, HbA1 c: 8.6 ± 1.3 %). The Taq I site was analysed by PCR followed by digestion with Taq I enzyme. Diabetic retinopathy was assessed by retinal angiography and classified as presence (n = 101) or absence (n = 99) of severe (preproliferative or proliferative) diabetic retinopathy. Results: Frequency of wild-type genotype TT was lower in patients with severe diabetic retinopathy (n = 27) when compared with control subjects (n = 42, OR = 0.5, p = 0.028). Allele frequencies were not different between patients (T: n = 112 and t: n = 90) and control subjects (T: n = 128, and t: n = 70, p = 0.075). Global χ2 (df = 2): p = 0.064. In subjects with diabetes duration of more than 25 years, TT was lower in severe diabetic retinopathy (n = 14) than control subjects (n = 18, OR = 0.3, p = 0.01). Allele frequencies were different between patients (T: n = 68 and t: n = 66) and control subjects (T: n = 52, OR = 0.5, and t: n = 26, OR = 1.9, p = 0.034). Global χ2 (df = 2): p = 0.024. In subjects with HbA1 c over 9 %, Tt was higher in patients (n = 28) than control subjects (n = 15, OR = 3.1, p = 0.019). Allele frequencies were not different between patients (T: n = 52 and t: n = 38) and control subjects (T: n = 57, and t: n = 29, p = 0.31). Global χ2 (df = 2): p = 0.035. Conclusion/interpretation: In French Type I (insulin-dependent) diabetic patients, we demonstrate an association between TT form (VDR) and low risk for severe diabetic retinopathy, especially in patients with long duration, and between Tt variant and high risk for severe diabetic retinopathy in subjects with poor glycaemic control. [Diabetologia (2002) 45: 436–442]