We assessed the day-to-day reproducible of vitreous fluorophotometry in 7 type 1 diabetic patients and in 1 healthy control subject. The coefficient of variation for duplicate measurements was 33.7, 40.3 and 23.6%, for the right eyes, the left eyes and mean values of both eyes, respectively. Assessment of reproducibility in 1 healthy subject, measured 4 times, resulted in coefficients of variation of about 60%. These values are somewhat worse than those obtained by others in healthy subjects. Since technical problems do not seem to play a significant role in this regard, our reproducibility results reflect the true biological variability of the permeability of the blood-retina barrier assessed by vitreous fluorophotometry in diabetic patients. This variability is substantial. Therefore, we conclude that this method is not precise enough to be used in intervention studies with a limited number of subjects.