Contrasting data about the association between proliferative diabetic retinopathy (PDR) and vitamin D status remain unknown. First, a hospital-based cross-sectional study consisting of 889 diabetic retinopathy (DR) and non-DR (NDR) patients was admitted. Further the accumulated evidence was performed to explore the association and dose-response relationship. Our study indicated that the odd ratio for PDR in vitamin D deficiency (VDD) individuals was significantly increased (1.60, 95 % CI 1.06-2.42), compared with NDR in vitamin D sufficiency individuals, adjusted by age, sex, diabetic duration, and HbA1c. Four studies plus our study with data on vitamin D levels in 4970 patients with PDR and NDR subjects are compared. Association between vitamin D deficiency and risk of PDR exists (OR = 1.69, 95 % CI 1.40-2.05; I-2 = 0 %, p = 0.61). Association between a nonlinear trend for vitamin D decrease with risk of DR was significant (chi(2) = 16.53, p = 0.0003). No significant heterogeneity in identified studies was found (goodness of fit chi(2) = 2.98, p = 0.225). It is concluded that vitamin D deficiency is significantly associated with risk of proliferative diabetic retinopathy.