Insulin administered in eyedrop from with a surfactant agent has been shown to be clinically effective in treating diabetes in animal models. Concentrations of insulin as high as 100 U/ml in saline were shown to produce no detectable clinical toxicity to human eyes in single-dose administration. We sought to investigate the local toxicity of insulin in human eyes during long-term, multidose administration. A prospective, randomized, placebo controlled, double-masked study was conducted involving eight healthy volunteers. Subjects were given 50 μl sterile saline containing 100 U/ml crystalline porcine insulin randomized to one eye and 50 μl placebo (sterile saline) to the fellow eye administered twice daily for 8 weeks. Subjective ocular irritation and visual acuity and objective assessment of the eyelids, conjunctiva, cornea, anterior chamber, crystalline lens, pupil size, and intraocular pressure were evaluated. Blood D-glucose levels were monitored to assess glycemic levels. There was no statistically significant difference (p>0.05) observed between insulin-treated and placebo-treated eyes. Eyedrops containing insulin were subjectively as comfortable and objectively as clinically innocuous as sterile saline alone. The results of this study demonstrate that insulin (100 U/ml) in saline is nontoxic to the human eye after long-term, multi-dose exposure. Topical administration of insulin combined with an absorption-promoting agent may be a practical and feasible alternative to multiple daily subcutaneous injections or implanted pump devices currently used in the long-term treatment of diabetes mellitus if a nonirritating absorption-promoting agent can be identified. Further studies are warranted to determine the toxicity and efficacy of these absorption-promoting agents in humans and to develop an eyedrop formulation that maximizes safety, efficacy and compliance in patients with diabetes mellitus. © 1994, Mary Ann Liebert, Inc. All rights reserved.