The purpose of this work was to determine the within-session variability and between-session repeatability of the Canon Laser Blood Flowmeter (CLBF), Model 100, an instrument that permits the noninvasive measurement of retinal blood flow. The CLBF calculates flow in microl/min based on the Poiseuille principle. One eye of 20 normal subjects (mean age 36.5; SD 9.7 years) was randomly chosen. A minimum of five measurements was acquired of a temporal arteriole approximately 1 disc diameter from the optic nerve. Measurements were repeated within a 1-month period. Blood pressure and intraocular pressure were measured. The coefficient of variation (COV) and the coefficient of repeatability (COR) were calculated for each individual. The individual COVs for diameter, velocity, and flow ranged from 0.5 to 6.5% (median 2.0%), 4.8 to 39.7% (median 19.9%), and 4.8 to 37.3% (median 19.3%), respectively. The group mean CORs for diameter, velocity, and flow were 5.2 microm (relative to a mean effect of 104.6 microm), 8.8 mm/s (relative to a mean effect of 33.9 mm/s), and 2.6 microl/min (relative to a mean effect of 8.8 microl/min), respectively. The CLBF gave consistent and repeatable measurements of blood flow within retinal arterioles in normal subjects. Given the range of individual variation in the velocity measurement, and thus flow, confidence limits for retinal hemodynamics need to be determined on an individual basis.