The Mechanical Engineering Laboratory of GEC Power Engineering Ltd. has developed a dual focus laser anemometer primarily for the measurement of flow velocities and directions in turbomachinery components. Typically, the anemometer is required to measure supersonic velocities in hostile environments with access to the point of measurement so restricted that a backscatter optical arrangement must be used. The GEC anemometer is an integrated optical unit which is mounted on a motorized traverse table thereby providing the opportunity for remote operation of the whole unit. Signals from the photomultipliers are processed in a Malvern correlator which is interfaced with a HP 9825A desk-top computer for on-line data reduction. In fact, the operation of the whole system is centred on this computer which is also used to control the motorized traverse movements. The policy within GEC is that experimental work should be directed towards the validation of design assumptions and the improvement of theoretical prediction techniques. To this end the anemometry equipment has recently been employed in the measurement of flow conditions through a cascade of turbine guide blades with supersonic exit velocities. The blades under test represented typical turbine sections and in this application the cascade was fed from a compressed air supply which exhausted from the cascade to atmosphere. The anemometer measurements were all made at the mid-span plane of the cascade where the flow conditions were two-dimensional, a series of traverses being conducted to evaluate flow conditions within and downstream of the blade passages. These measurements also permitted a detailed investigation of shock waves and blade wakes, features which had been previously located from Schlieren photographs. A programme of theoretical predictions of the flow through the cascade was subsequently undertaken using a two-dimensional time-marching method. The comparison between these predictions and the test data revealed serious errors in the flow predictions near the blade trailing edges, the errors being attributed to difficulties in accurately modelling the blade trailing edge geometry in the computer program. These problems are being attended to during the course of program development, which has clearly benefited from the availability of accurate experimental data.