We treated 69 eyes of 44 patients who had clinically significant macular edema with argon or krypton laser photocoagulation or both. The mean initial visual acuity was 20/40. Patients were followed for 6 to 63 (mean 19.7 +/- 11.0) months. There was a mean loss of three letters in acuity, but the difference between initial and final acuity was not statistically significant. In 43 eyes the acuity remained within two lines of the initial acuity, in 17 there was a loss of two lines or more, and in 9 the acuity improved by two lines or more. The initial acuity correlated highly to the final acuity (p less than 0.0001). Multiple linear regression showed that pretreatment existence of macular ischemia or diffuse macular edema or both, when adjusted for initial acuity, age, insulin dependence and duration of diabetes mellitus, was predictive of poorer final acuity than in patients without ischemia or diffuse edema. The results support a consensus that macular edema should be treated early. In view of the conflicting experimental and clinical studies of grid techniques of photocoagulation and the role of the krypton laser, further clinical studies of the treatment of diabetic macular edema are warranted.