This paper introduces a new instrument for measuring groundwater velocity in a borehole using a laser Doppler velocimeter (LDV) technique. The major improvement of the proposed LDV system is that the probe can be inserted into a borehole that has a minimum diameter of 10 cm and that a disk-type rotating grating is used as a frequency shifter for low velocities in order to determine the velocity sign. The speed of the glass plate passing through the LDV measurement point is calibrated to demonstrate the accuracy of the new LDV system. From a comparison with the true constant speed of the glass plate, we confirm that the new LDV system can measure low speeds in the range from 10 -5 to 10--' cm/s at a measurement error of less than 8%. Laboratory experiments were carded out to investigate the performance of the new LDV system for the horizontal velocity measurement of groundwater in a borehole. The result obtained by the proposed LDV system shows that the horizontal velocity at the center of an unscreened borehole is 3 times larger than the seepage velocity in a confined aquifer. This result agrees with the theoretical solution for groundwater flow in a borehole derived by Sano (1983). The validity of the theoretical solution is experimentally verified in the laboratory using the current LDV system. Although the detectable minimum velocity in a borehole is limited by the velocity fluctuations due to the BrownJan motion of scattering particles, results obtained in the experiments using 0.804-/zm polystyrene particles show that the LDV system described in this paper can measure with a sufficient accuracy both the magnitude and the direction of groundwater velocity down to 10 -4 cm/s in a borehole. INTRODUCTION When groundwater velocity is investigated by injecting a tracer such as a radioactive isotope, dye, or salt into a borehole and the arrival time of the tracer at observation wells arranged a few meters away from the injection well is observed, there are four fundamental problems. First, several observation wells will have to be installed around the injection well if the flow direction is unknown. Second, it is difficult to accurately evaluate the arrival time due to effects of dispersion, diffusion, chemical reaction, and adsorption of the tracer. Third, injection of a large amount of tracer will cause groundwater pollution and a local change in the groundwater movement. Fourth, long-time observations are usually required because the velocity is usually very small. In recent years, the point dilution method [Drost et al., 1968] has been commonly used for groundwater flow measurements. In the point dilution method the velocity measurement is carried out in a single borehole, and the first two obstacles can be eliminated. However, since this method requires a radioisotopic tracer, the third problem still remains. Also, the effects of adsorption and density convection currents need to be carefully considered. Furthermore, as mentioned by Drost et al. [1968], the measuring period by the point dilution method is abou...