PURPOSE: To determine the incidence of bacterial contamination of needles used for intravitreal injections. METHODS: Patients undergoing intravitreal injections were enrolled prospectively. No pre-injection antibiotics were administered. Following povidone-iodine irrigation, conjunctival cultures were taken and the injection was performed. The needle was cultured. A dry control needle was exposed to the surgical field and cultured. RESULTS: No patients developed endophthalmitis. Eighteen injection needles (18%) yielded positive bacterial growth. The most commonly encountered organisms were Propionibacterium acnes (n = 8) and Staphylococcus epidermidis (n = 6). Four control needles showed positive growth, in 2 cases with the same organism as a matching positive used needle. The difference between contamination rates of used and control needles was significant (p = .002, McNemar's test). CONCLUSIONS: Bacterial contaminants are present on a substantial proportion of needles. Since the needle contacts both the ocular surface and the vitreous, it is possible that inoculation of the vitreous cavity occurs in such cases.