The two techniques, vibratory monitoring and pit counting, show promising results for the prediction of prototype cavitation erosion from model tests of Francis turbines. These cavitation detection methods are compared in two set-ups at different power levels, a laboratory high velocity cavitation jet and a full scale 270 MW Francis turbine. Excellent quantitative correlations are obtained in the jet tests between erosion rate, volume pitting rate measured on polished metal surfaces of different hardness with a laser profilometer and the mean square value of forces on the eroded specimen inferred from measurements with a high frequency accelerometer on the specimen holder. On the large prototype good coherent results are also obtained but the vibratory information requires much finer analysis. In particular the varying erosive cavitation intensity with power output level is well detected by both methods. The two cavitation detection techniques exhibit great dynamic range and can prove very useful in characterizing the erosive aggressiveness of cavitating flow both in large machines and in reduced scale models.