Purpose: To determine the intravisit and intervisit reproducibility of optical coherence tomography angiography measurements of macular vessel density in eyes with and without retinal diseases. Methods: Fifteen healthy volunteers and 22 patients with retinal diseases underwent repeated optical coherence tomography angiography (Angiovue Imaging System, Optovue Inc) scans after pupil dilation on 2 separate visit days. For each visit day, the eyes were scanned twice. Vessel density defined as the proportion of vessel area with flowing blood over the total measurement area was calculated using Angiovue software. Intravisit and intervisit reproducibility were summarized as coefficient of variations and intraclass correlation coefficients were calculated from variance component models. Results: The coefficient of variations representing the intravisit reproducibility of the superficial macular vessel density measurements for different quadrants on 3 mm x 3-mm scans varied from 2.1% to 4.9% and 3.4% to 6.8% for healthy and diseased eyes, respectively, and for the intervisit it was 2.9% to 5.1% and 4.0% to 6.8%, respectively. The coefficient of variations were lower in healthy eyes than in diseased eyes, lower for intravisit than for intervisit, lower on 3 mm x 3-mm scans than on 6 mm x 6-mm scans, and lower for paracentral subfields than for central subfield. Conclusion: The evidence presented here demonstrates good reproducibility of optical coherence tomography angiography for measurement of superficial macula vessel density in both healthy eyes and eyes with diabetic retinopathy without diabetic macular edema.