This study modified the microspheres method by optimizing the dose and size of microspheres (MS) to enable accurate ocular blood flow measurement in rats. Fluorescent MS, either 6, 8, 10 or 15mum diameter, were administered into the left ventricle of anesthetized adult Brown Norway rat in a dose of either 10(6), 5x10(6), or 10(7). The total number of MS entrapped in retina, choroid and optic nerve (N(tissue)) was quantified and compared between size and dose groups. The MS distribution in the retina and their reentry into systemic circulation were evaluated for different sized MS. The results showed that at the 5x10(6) dose, the N(tissue) of 8mum MS was significantly more than either 6 or 10mum MS in the retina (P<0.02) and optic nerve (P<0.03). The 10mum MS produced the highest N(tissue) for the choroid, as compared with either 8 or 6mum MS (P<0.03). At the 10(6) dose, no difference of N(tissue) was found between 8, 10, and 15mum MS in the retina. The 10mum MS yielded the highest N(tissue) in the choroid as compared to 8 and 15mum MS (P<0.003). The N(tissue) for 8mum MS was higher than both 10 and 15mum (P<0.01) MS in the optic nerve. No MS (>/=8mum) reentered the systemic circulation. The 15mum MS tended to lodge in pre-capillary arterioles and caused significant blood pressure increase during the injection. The blood flow measured with the optimal size MS (mean+/-SE) were 19+/-3.4 and 170+/-35mul/min in the retina and choroid, respectively; and 0.18+/-0.03mul/min per mm optic nerve. It is concluded that the 8mum MS are the optimal size for both retinal and optic nerve blood flow estimation; the 10mum for the choroid. The optimal dose for the retina was approximately 2.5x10(6), 0.5x10(6) for the choroid, and 5x10(6) approximately 10(7) for the optic nerve. The 15mum MS are inappropriate for ocular blood flow measurements in rats.