Two compounds, bromovinyldeoxyuridine ([E]-5-[2-bromovinyl]-2'-deoxyuridine) and trifluridine (5-trifluoromethyl-2' deoxyuridine) were compared for their efficacy in the topical treatment of experimental stroma herpetic keratitis produced by the injection of live herpes simplex virus, type 1, into the corneal stroma of rabbits. The trifluridine was used as 1% eyedrops, whereas bromovinyldeoxyuridine was used as either 0.1% or 0.5% eyedrops. All three treatment regimens caused a substantial healing of stroma disease in comparison with placebo treatment. The bromovinyldeoxyuridine (whether used as 0.1% or 0.5%) proved superior to 1% trifluridine eyedrops when the treatment was started one day after virus inoculation. However, bromovinyldeoxyuridine and trifluridine were equally effective if the treatment was started seven days after infection. The 0.5% bromovinyldeoxyuridine eyedrops seemed to be more effective in controlling the complicating severe iritis and secondary glaucoma than either 0.1% bromovinyldeoxyuridine or 1% trifluridine. No toxic effects were observed with bromovinyldeoxyuridine in any eye, whereas 1% trifluridine produced punctate epitheliopathy in some eyes.