PURPOSE. Structural retinal microvascular changes have been identified as risk markers of diabetic retinopathy (DR). In order to estimate the retinal response of neuroprotective eye drops, we aimed to evaluate the effect of topical retinal neuroprotection on retinal microvascular changes in early DR. METHODS. Patients with type 2 diabetes with no or early DR were randomized 1:1:1 to topical treatment with placebo, brimonidine, or somatostatin in a 96-week prospective, phase II to III, European multicenter trial. Retinal vascular calibers were measured semiautomatically in digital fundus images by certified graders at baseline and follow-up and summarized as central retinal arteriolar and venular equivalent (CRAE and CRVE). RESULTS. Of 449 patients originally included, 297 completed the study with gradable retinal images. Median age and duration of diabetes was 64.5 and 9.9 years, and 65.7% were male. At baseline, Early Treatment Diabetic Retinopathy Study levels were 10 (no DR, 42.8%), 20 (minimal DR, 28.3%), and 35 (mild DR, 29.0%), and CRAE and CRVE did not differ between groups. As opposed to patients with no or minimal DR at baseline, patients with mild DR in the active groups developed a larger retinal arteriolar (brimonidine: +6.2 µm, P = 0.006; somatostatin: +7.2 µm, P = 0.006) and venular (brimonidine: +13.9 µm, P = 0.01; somatostatin: +14.3 µm, P = 0.0001) caliber in contrast to those in the placebo group. CONCLUSIONS. Topical treatment with brimonidine and somatostatin causes retinal arteriolar and venular dilation in patients with type 2 diabetes and preexisting early DR. Upcoming studies should elaborate on the potential of these findings in arresting early DR.