To examine the relationship of optical coherence tomography (OCT) measured macular thickness to retinopathy severity in patients with diabetes and no clinically detectable macular edema. Retrospective observational case series. Three hundred eighty-three eyes of 383 patients of a private retina practice; including 100 normal eyes of patients without diabetes, 100 eyes of diabetics without retinopathy, 100 eyes of diabetics with mild to moderate retinopathy, 35 eyes of diabetics with severe nonproliferative or proliferative retinopathy, and 48 eyes of diabetics with regressed proliferative retinopathy. Review of clinical charts and optical coherence tomography measurements. Central subfield mean thickness (CSMT), inner and outer zone measurements, and total macular volume. Central subfield mean thicknesses (mean +/- standard deviation) were 208+/-22, 198+/-25, 204+/-26, 224+/-38, and 205+/-27 microm for normals, eyes of diabetics without retinopathy, eyes with mild to moderate nonproliferative retinopathy, eyes with severe nonproliferative to proliferative retinopathy, and eyes with regressed proliferative retinopathy, respectively. For all groups, mean CSMT was larger in males than in females. Statistically significant differences by gender were observed for normals, diabetics without retinopathy, and diabetics with mild to moderate nonproliferative retinopathy (mean differences, 12, 14, and 18 microm, respectively; Ps = 0.0057, 0.0057, and 0.0002). For increasing retinopathy severity, the probability of macular thickening detected by OCT but not detected by clinical examination increased. Fifteen percent of eyes with severe nonproliferative or proliferative retinopathy and no clinically detected edema had OCT-measured macular thickening. Because OCT measurements are gender dependent, gender balance or statistical adjustment for gender imbalances of compared groups in OCT studies of diabetic macular edema is important. As retinopathy severity increases, the probability of subclinical edema rises. Except for an individual baseline measurement possibly useful for longitudinal comparison, the data suggest that there is little reason routinely to obtain OCT in eyes with diabetes and no retinopathy or mild to moderate diabetic retinopathy when clinical examination fails to show macular edema.