Objective: To investigate the efficacy and adverse reactions of 5% povidone-iodine in removing bacteria from the conjunctival sac with different durations. Methods: Randomized controlled study. A total of 420 patients who underwent cataract surgery in Daping Hospital, Army Medical University from December 2017 to June 2018 were selected. Non-surgical eyes (420 eyes) were selected as the study subjects and divided into 4 groups randomly: 30-second group, 1.0-min group, 2.0-min group and 3.5-min group. On the day of surgery, domestic 5% povidone-iodine was used to flush the conjunctival sac for 30 seconds, 1.0 min, 2.0 min and 3.5 min, respectively. The conjunctival sac specimens were collected for bacterial culture and bacterial identification before and after flushing the conjunctival sac with povidone-iodine. The positive rates of bacterial culture and bacterial growth were compared. The patients' ocular surface was observed and the incidence of corneal epithelial injury was recorded at 1 hour and 1 day after surgery. The positive rates of bacterial culture and corneal epithelial injury between groups were compared by Pearson chi-square test. Results: After excluding 20 patients with suspected specimens contamination, 400 patients (400 non-surgical eyes) were enrolled, including 191 males and 209 females, with an average age of 66.8 years. Before flushing the conjunctival sac, the positive rates of bacterial culture in the 30-second group, 1.0-min group, 2.0-min group and 3.5-min group were 44.8% (43/96), 43.3% (39/90), 43.1% (47/109) and 43.8% (46/105), respectively, with no statistically significant difference (χ(2)=0.066, P=0.996). After flushing, the positive rates of conjunctival sac bacterial culture in the 4 groups were 29.2% (28/96), 31.1% (28/90), 13.8% (15/109) and 13.3% (14/105), respectively. The differences between the 30-second group and 2.0-min group (χ(2)=7.308, P=0.007), between the 1.0-min group and 2.0-min group (χ(2)=8.760, P=0.003), between the 30-second group and 3.5-min group (χ(2)=7.606, P=0.006), and between the 1.0-min group and 3.5-min group (χ(2)=9.063, P=0.003) were statistically significant. At 1 hour after surgery, mild corneal epithelial injury occurred in each group, with a rate of 16.7% (16/96), 18.9% (17/90), 20.2% (22/109) and 34.3% (36/105), respectively. The differences between the 30-second group and 3.5-min group (χ(2)=8.118, P=0.004), between the 1.0-min group and 3.5-min group (χ(2)=5.804, P=0.016), and between the 2.0-min group and 3.5-min group (χ(2)=5.383, P=0.020) were statistically significant. At 1 day after surgery, there was no occurrence of new injury, and the incidence of mild corneal injury in each group was 3.1% (3/96), 5.6% (5/90), 9.2% (10/109) and 15.2% (16/105), respectively. There was statistically significant difference between the 30-second group and 3.5-min group (χ(2)=8.597, P=0.003), and between the 1.0-min group and 3.5-min group (χ(2)=4.728, P=0.030). The corneal epithelial injury healed completely at 1 week after surgery. Conclusions: The preoperative bacterial load of the conjunctival sac is more effectively reduced with 5% povidone-iodine in the 2.0-min and 3.5-min than in the 30-second and 1.0-min, and the 2-min is superior to the 3.5-min in the occurrence of corneal epithelial injury at 1 hour after surgery. Irrigation of the conjunctival sac with 5% povidone-iodine for 2 min is effective and safe, which can be an alternative measure. (Chin J Ophthalmol, 2019, 55: 509-514).