BACKGROUND Diabetic retinopathy can result in impaired photostress recovery time despite normal visual acuity and fundoscopic appearance. The Macular Degeneration Detector (MDD-2) is a novel flash photostress recovery time device. In this study, we examine the repeatability of the MDD-2 in normal and diabetic subjects. METHODS One hundred and ninety one (90 women, 101 men) subjects were recruited and divided into 1 of the 3 study groups (normal controls, n = 40; diabetes no retinopathy, n = 98; nonproliferative diabetic retinopathy, n = 53). Photostress recovery time was measured three times in the study eye using the MDD-2, each measurement separated by a 5-minute interval. RESULTS Repeated measures analysis of variance revealed no statistically significant learning or fatigue effects on intrameasurement repeatability for any group. Photostress recovery time measures were broadly similar and typically not statistically significantly different between study groups. The coefficient of repeatability reached clinically acceptable levels once the initial photostress recovery time measure, which demonstrated increased variability and latency compared with all subsequent measures, was excluded. CONCLUSION The MDD-2 seems to provide repeatable photostress recovery time measurements among naive diabetic subjects. The device does not, however, seem capable of differentiating normal and nonproliferative diabetic eyes, and would not be suitable for inclusion in diabetic retinopathy screening protocol.